Install detectron2

from google.colab import drive
drive.mount('/content/drive')
data_dir = "/content/drive/MyDrive/MIT/3. Spring 2022/6.869/6.869 Project/v-coco"

# %cd /content/drive/MyDrive/MIT/3.\ Spring\ 2022/6.869/6.869\ Project/v-coco/

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).

%cd /content/drive/.shortcut-targets-by-id/1Eg2dTJ9lloHImzZiTfOZ7_jOP1sTh--c/6.869 Project/v-coco/

/content/drive/.shortcut-targets-by-id/1Eg2dTJ9lloHImzZiTfOZ7_jOP1sTh--c/6.869 Project/v-coco

!pip install -U torch torchvision
!pip install git+https://github.com/facebookresearch/fvcore.git
import torch, torchvision
torch.__version__

Requirement already satisfied: torch in /usr/local/lib/python3.7/dist-packages (1.11.0+cu113)
Requirement already satisfied: torchvision in /usr/local/lib/python3.7/dist-packages (0.12.0+cu113)
Requirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from torch) (4.2.0)
Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in /usr/local/lib/python3.7/dist-packages (from torchvision) (7.1.2)
Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from torchvision) (1.21.6)
Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from torchvision) (2.23.0)
Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision) (3.0.4)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision) (2021.10.8)
Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision) (1.24.3)
Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision) (2.10)
Collecting git+https://github.com/facebookresearch/fvcore.git
  Cloning https://github.com/facebookresearch/fvcore.git to /tmp/pip-req-build-1zxl_lz6
  Running command git clone -q https://github.com/facebookresearch/fvcore.git /tmp/pip-req-build-1zxl_lz6
Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from fvcore==0.1.5) (1.21.6)
Collecting yacs>=0.1.6
  Downloading yacs-0.1.8-py3-none-any.whl (14 kB)
Collecting pyyaml>=5.1
  Downloading PyYAML-6.0-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_12_x86_64.manylinux2010_x86_64.whl (596 kB)
     |████████████████████████████████| 596 kB 4.3 MB/s 
Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from fvcore==0.1.5) (4.64.0)
Requirement already satisfied: termcolor>=1.1 in /usr/local/lib/python3.7/dist-packages (from fvcore==0.1.5) (1.1.0)
Requirement already satisfied: Pillow in /usr/local/lib/python3.7/dist-packages (from fvcore==0.1.5) (7.1.2)
Requirement already satisfied: tabulate in /usr/local/lib/python3.7/dist-packages (from fvcore==0.1.5) (0.8.9)
Collecting iopath>=0.1.7
  Downloading iopath-0.1.9-py3-none-any.whl (27 kB)
Collecting portalocker
  Downloading portalocker-2.4.0-py2.py3-none-any.whl (16 kB)
Building wheels for collected packages: fvcore
  Building wheel for fvcore (setup.py) ... done
  Created wheel for fvcore: filename=fvcore-0.1.5-py3-none-any.whl size=65172 sha256=4dd2c48a44b3f6088e98d5a602739f54f755f9e7149f476bda211083063457d4
  Stored in directory: /tmp/pip-ephem-wheel-cache-wz77symq/wheels/24/1d/09/8167de727fe5b74f832b6fcb5d9069d8f03ca29f337bfe484d
Successfully built fvcore
Installing collected packages: pyyaml, portalocker, yacs, iopath, fvcore
  Attempting uninstall: pyyaml
    Found existing installation: PyYAML 3.13
    Uninstalling PyYAML-3.13:
      Successfully uninstalled PyYAML-3.13
Successfully installed fvcore-0.1.5 iopath-0.1.9 portalocker-2.4.0 pyyaml-6.0 yacs-0.1.8

'1.11.0+cu113'

!git clone https://github.com/facebookresearch/detectron2 detectron2_repo
!pip install -e detectron2_repo

fatal: destination path 'detectron2_repo' already exists and is not an empty directory.
Obtaining file:///content/drive/.shortcut-targets-by-id/1Eg2dTJ9lloHImzZiTfOZ7_jOP1sTh--c/6.869%20Project/v-coco/detectron2_repo
Requirement already satisfied: Pillow>=7.1 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (7.1.2)
Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (3.2.2)
Requirement already satisfied: pycocotools>=2.0.2 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (2.0.4)
Requirement already satisfied: termcolor>=1.1 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (1.1.0)
Requirement already satisfied: yacs>=0.1.8 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (0.1.8)
Requirement already satisfied: tabulate in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (0.8.9)
Requirement already satisfied: cloudpickle in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (1.3.0)
Requirement already satisfied: tqdm>4.29.0 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (4.64.0)
Requirement already satisfied: tensorboard in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (2.8.0)
Requirement already satisfied: fvcore<0.1.6,>=0.1.5 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (0.1.5)
Requirement already satisfied: iopath<0.1.10,>=0.1.7 in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (0.1.9)
Requirement already satisfied: future in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (0.16.0)
Requirement already satisfied: pydot in /usr/local/lib/python3.7/dist-packages (from detectron2==0.6) (1.3.0)
Collecting omegaconf>=2.1
  Downloading omegaconf-2.1.2-py3-none-any.whl (74 kB)
     |████████████████████████████████| 74 kB 2.0 MB/s 
Collecting hydra-core>=1.1
  Downloading hydra_core-1.1.2-py3-none-any.whl (147 kB)
     |████████████████████████████████| 147 kB 26.9 MB/s 
Collecting black==21.4b2
  Downloading black-21.4b2-py3-none-any.whl (130 kB)
     |████████████████████████████████| 130 kB 47.1 MB/s 
Collecting scipy>1.5.1
  Downloading scipy-1.7.3-cp37-cp37m-manylinux_2_12_x86_64.manylinux2010_x86_64.whl (38.1 MB)
     |████████████████████████████████| 38.1 MB 1.3 MB/s 
Collecting toml>=0.10.1
  Downloading toml-0.10.2-py2.py3-none-any.whl (16 kB)
Collecting mypy-extensions>=0.4.3
  Downloading mypy_extensions-0.4.3-py2.py3-none-any.whl (4.5 kB)
Collecting pathspec<1,>=0.8.1
  Downloading pathspec-0.9.0-py2.py3-none-any.whl (31 kB)
Collecting regex>=2020.1.8
  Downloading regex-2022.4.24-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (749 kB)
     |████████████████████████████████| 749 kB 47.5 MB/s 
Requirement already satisfied: typing-extensions>=3.7.4 in /usr/local/lib/python3.7/dist-packages (from black==21.4b2->detectron2==0.6) (4.2.0)
Requirement already satisfied: click>=7.1.2 in /usr/local/lib/python3.7/dist-packages (from black==21.4b2->detectron2==0.6) (7.1.2)
Requirement already satisfied: appdirs in /usr/local/lib/python3.7/dist-packages (from black==21.4b2->detectron2==0.6) (1.4.4)
Collecting typed-ast>=1.4.2
  Downloading typed_ast-1.5.3-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_12_x86_64.manylinux2010_x86_64.whl (843 kB)
     |████████████████████████████████| 843 kB 53.3 MB/s 
Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.7/dist-packages (from fvcore<0.1.6,>=0.1.5->detectron2==0.6) (6.0)
Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from fvcore<0.1.6,>=0.1.5->detectron2==0.6) (1.21.6)
Collecting importlib-resources<5.3
  Downloading importlib_resources-5.2.3-py3-none-any.whl (27 kB)
Collecting antlr4-python3-runtime==4.8
  Downloading antlr4-python3-runtime-4.8.tar.gz (112 kB)
     |████████████████████████████████| 112 kB 60.5 MB/s 
Requirement already satisfied: zipp>=3.1.0 in /usr/local/lib/python3.7/dist-packages (from importlib-resources<5.3->hydra-core>=1.1->detectron2==0.6) (3.8.0)
Requirement already satisfied: portalocker in /usr/local/lib/python3.7/dist-packages (from iopath<0.1.10,>=0.1.7->detectron2==0.6) (2.4.0)
Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->detectron2==0.6) (3.0.8)
Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib->detectron2==0.6) (0.11.0)
Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->detectron2==0.6) (1.4.2)
Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->detectron2==0.6) (2.8.2)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.1->matplotlib->detectron2==0.6) (1.15.0)
Requirement already satisfied: grpcio>=1.24.3 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (1.44.0)
Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (0.4.6)
Requirement already satisfied: absl-py>=0.4 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (1.0.0)
Requirement already satisfied: protobuf>=3.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (3.17.3)
Requirement already satisfied: setuptools>=41.0.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (57.4.0)
Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (1.8.1)
Requirement already satisfied: wheel>=0.26 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (0.37.1)
Requirement already satisfied: requests<3,>=2.21.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (2.23.0)
Requirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (3.3.6)
Requirement already satisfied: tensorboard-data-server<0.7.0,>=0.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (0.6.1)
Requirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (1.0.1)
Requirement already satisfied: google-auth<3,>=1.6.3 in /usr/local/lib/python3.7/dist-packages (from tensorboard->detectron2==0.6) (1.35.0)
Requirement already satisfied: cachetools<5.0,>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard->detectron2==0.6) (4.2.4)
Requirement already satisfied: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard->detectron2==0.6) (0.2.8)
Requirement already satisfied: rsa<5,>=3.1.4 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard->detectron2==0.6) (4.8)
Requirement already satisfied: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.7/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard->detectron2==0.6) (1.3.1)
Requirement already satisfied: importlib-metadata>=4.4 in /usr/local/lib/python3.7/dist-packages (from markdown>=2.6.8->tensorboard->detectron2==0.6) (4.11.3)
Requirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.7/dist-packages (from pyasn1-modules>=0.2.1->google-auth<3,>=1.6.3->tensorboard->detectron2==0.6) (0.4.8)
Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard->detectron2==0.6) (1.24.3)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard->detectron2==0.6) (2021.10.8)
Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard->detectron2==0.6) (2.10)
Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard->detectron2==0.6) (3.0.4)
Requirement already satisfied: oauthlib>=3.0.0 in /usr/local/lib/python3.7/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard->detectron2==0.6) (3.2.0)
Building wheels for collected packages: antlr4-python3-runtime
  Building wheel for antlr4-python3-runtime (setup.py) ... done
  Created wheel for antlr4-python3-runtime: filename=antlr4_python3_runtime-4.8-py3-none-any.whl size=141230 sha256=81281dda8dbbdbb02c852c1b208a2d3197969afa54ef1e649e9a62f044948d33
  Stored in directory: /root/.cache/pip/wheels/ca/33/b7/336836125fc9bb4ceaa4376d8abca10ca8bc84ddc824baea6c
Successfully built antlr4-python3-runtime
Installing collected packages: antlr4-python3-runtime, typed-ast, toml, regex, pathspec, omegaconf, mypy-extensions, importlib-resources, scipy, hydra-core, black, detectron2
  Attempting uninstall: regex
    Found existing installation: regex 2019.12.20
    Uninstalling regex-2019.12.20:
      Successfully uninstalled regex-2019.12.20
  Attempting uninstall: importlib-resources
    Found existing installation: importlib-resources 5.7.1
    Uninstalling importlib-resources-5.7.1:
      Successfully uninstalled importlib-resources-5.7.1
  Attempting uninstall: scipy
    Found existing installation: scipy 1.4.1
    Uninstalling scipy-1.4.1:
      Successfully uninstalled scipy-1.4.1
  Running setup.py develop for detectron2
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
albumentations 0.1.12 requires imgaug<0.2.7,>=0.2.5, but you have imgaug 0.2.9 which is incompatible.
Successfully installed antlr4-python3-runtime-4.8 black-21.4b2 detectron2-0.6 hydra-core-1.1.2 importlib-resources-5.2.3 mypy-extensions-0.4.3 omegaconf-2.1.2 pathspec-0.9.0 regex-2022.4.24 scipy-1.7.3 toml-0.10.2 typed-ast-1.5.3

Train on a custom COCO dataset

# Some basic setup
import vsrl_utils as vu
import vsrl_eval as ve
import json

# Setup detectron2 logger
import detectron2
from detectron2.utils.logger import setup_logger
setup_logger()

# import some common libraries
import matplotlib.pyplot as plt
import numpy as np
import cv2
from google.colab.patches import cv2_imshow

# import some common detectron2 utilities
from detectron2.engine import DefaultPredictor
from detectron2.config import get_cfg
from detectron2.utils.visualizer import Visualizer
from detectron2.data import MetadataCatalog, DatasetCatalog

Register the V-COCO dataset to detectron2.

# Load COCO annotations for V-COCO images
coco = vu.load_coco()

# Load the VCOCO annotations for vcoco_train image set
vcoco_all = vu.load_vcoco('vcoco_val')
for x in vcoco_all: 
    x = vu.attach_gt_boxes(x, coco)
increment = 0

!python script_pick_annotations_val_only.py coco/annotations
!python script_pick_annotations_train_only.py coco/annotations

coco/annotations, data
Loading training annotations from coco/annotations/instances_train2014.json
Loading validating annotations from coco/annotations/instances_val2014.json
Writing COCO annotations needed for V-COCO to data/instances_vcoco_val_2014_t.json.
coco/annotations, data
Loading training annotations from coco/annotations/instances_train2014.json
Loading validating annotations from coco/annotations/instances_val2014.json
Writing COCO annotations needed for V-COCO to data/instances_vcoco_train_2014.json.

relevant_classes = ["hold", "carry", "point", "eat", "drink", "stand", "talk_on_phone"] #["hold", "look", "carry", "point", "hit", "eat", "drink", "stand", "talk_on_phone"]
classes = [x['action_name'] for x in vcoco_all]

final_json = {"images": [], "annotations": [], "categories": []}
all_ids = []
for category, action in enumerate(relevant_classes):
  cls_id = classes.index(action)
  vcoco = vcoco_all[cls_id]
  positive_index = np.where(vcoco['label'] == 1)[0]
  image_ids = vcoco['image_id'][positive_index][:,0]
  anno_ids = vcoco['ann_id'][positive_index][:,0]

  final_data = {}
  final_data["images"] = coco.loadImgs(ids=image_ids)
  action_annos = coco.loadAnns(ids=anno_ids)[:]
  new_annos = []
  for a_anno in action_annos:
    new_anno = {**a_anno}
    new_anno["category_id"] = category + 1
    new_anno["id"] = int(str((category+1)*10000000000) + str(a_anno["id"]))
    new_annos.append(new_anno)
    # if a_anno["id"] in all_ids: print(a_anno["id"])
    all_ids.append(new_anno["id"])
  # print(int(str((category+1)*10000000000) + str(a_anno["id"])))
  final_data["annotations"] = new_annos
  final_data["categories"] = [{"supercategory": action, "id": category+1, "name": action}]

  for key in final_json.keys():
    final_json[key] = [*final_json[key], *final_data[key]]

increment = increment + 1
with open("./data/instances_vcoco_actions_train.json", 'w') as fp:
    json.dump(final_json, fp)

from detectron2.data.datasets import register_coco_instances
register_coco_instances("rel_actions"+str(increment), {}, "./data/instances_vcoco_actions_train.json", "./coco/train2014")
print("rel_actions"+str(increment))

rel_actions1

train_metadata = MetadataCatalog.get("rel_actions"+str(increment))
train_dicts = DatasetCatalog.get("rel_actions"+str(increment))

[05/08 23:31:46 d2.data.datasets.coco]: Loaded 1951 images in COCO format from ./data/instances_vcoco_actions_train.json

increment = increment + 1
with open("./data/instances_vcoco_actions_val_2014.json", 'w') as fp:
    json.dump(final_json, fp)

from detectron2.data.datasets import register_coco_instances
register_coco_instances("rel_actions_val"+str(increment), {}, "./data/instances_vcoco_actions_val_2014.json", "./coco/train2014")
print("rel_actions_val"+str(increment))

val_metadata = MetadataCatalog.get("rel_actions_val"+str(increment))
val_dicts = DatasetCatalog.get("rel_actions_val"+str(increment))

rel_actions_val3
[05/09 00:13:52 d2.data.datasets.coco]: Loaded 2215 images in COCO format from ./data/instances_vcoco_actions_val_2014.json

To verify the data loading is correct, let's visualize the annotations of randomly selected samples in the training set:

import random

for d in random.sample(train_dicts, 3):
    # print(d)
    img = cv2.imread(d["file_name"]) #Start at 7 due to script name mismatch
    visualizer = Visualizer(img[:, :, ::-1], metadata=train_metadata, scale=0.5)
    vis = visualizer.draw_dataset_dict(d)
    cv2_imshow(vis.get_image()[:, :, ::-1])

Now, let's fine-tune a coco-pretrained R50-FPN Mask R-CNN model on the fruits_nuts dataset. It takes ~6 minutes to train 300 iterations on Colab's K80 GPU.

from detectron2.engine import DefaultTrainer
from detectron2.config import get_cfg
import os

cfg = get_cfg()
cfg.merge_from_file("./detectron2_repo/configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml")
cfg.DATASETS.TRAIN = ("rel_actions1",)
cfg.DATASETS.TEST = ()   # no metrics implemented for this dataset
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = "detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl"  # initialize from model zoo
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.SOLVER.BASE_LR = 0.002
cfg.SOLVER.MAX_ITER = 1000    
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 128   # faster, and good enough for this toy dataset
cfg.MODEL.ROI_HEADS.NUM_CLASSES = len(relevant_classes)  # all relevant classes
cfg.OUTPUT_DIR = "./outputs/action_only_detector"

os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
trainer = DefaultTrainer(cfg)
trainer.resume_or_load(resume=False)
trainer.train()

[05/08 23:44:11 d2.engine.defaults]: Model:
GeneralizedRCNN(
  (backbone): FPN(
    (fpn_lateral2): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral3): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral4): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output4): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral5): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output5): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (top_block): LastLevelMaxPool()
    (bottom_up): ResNet(
      (stem): BasicStem(
        (conv1): Conv2d(
          3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False
          (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
        )
      )
      (res2): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv1): Conv2d(
            64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
      )
      (res3): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv1): Conv2d(
            256, 128, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
      )
      (res4): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
          (conv1): Conv2d(
            512, 256, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (4): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (5): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
      )
      (res5): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
          (conv1): Conv2d(
            1024, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
      )
    )
  )
  (proposal_generator): RPN(
    (rpn_head): StandardRPNHead(
      (conv): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (objectness_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (anchor_deltas): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
    (anchor_generator): DefaultAnchorGenerator(
      (cell_anchors): BufferList()
    )
  )
  (roi_heads): StandardROIHeads(
    (box_pooler): ROIPooler(
      (level_poolers): ModuleList(
        (0): ROIAlign(output_size=(7, 7), spatial_scale=0.25, sampling_ratio=0, aligned=True)
        (1): ROIAlign(output_size=(7, 7), spatial_scale=0.125, sampling_ratio=0, aligned=True)
        (2): ROIAlign(output_size=(7, 7), spatial_scale=0.0625, sampling_ratio=0, aligned=True)
        (3): ROIAlign(output_size=(7, 7), spatial_scale=0.03125, sampling_ratio=0, aligned=True)
      )
    )
    (box_head): FastRCNNConvFCHead(
      (flatten): Flatten(start_dim=1, end_dim=-1)
      (fc1): Linear(in_features=12544, out_features=1024, bias=True)
      (fc_relu1): ReLU()
      (fc2): Linear(in_features=1024, out_features=1024, bias=True)
      (fc_relu2): ReLU()
    )
    (box_predictor): FastRCNNOutputLayers(
      (cls_score): Linear(in_features=1024, out_features=8, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=28, bias=True)
    )
    (mask_pooler): ROIPooler(
      (level_poolers): ModuleList(
        (0): ROIAlign(output_size=(14, 14), spatial_scale=0.25, sampling_ratio=0, aligned=True)
        (1): ROIAlign(output_size=(14, 14), spatial_scale=0.125, sampling_ratio=0, aligned=True)
        (2): ROIAlign(output_size=(14, 14), spatial_scale=0.0625, sampling_ratio=0, aligned=True)
        (3): ROIAlign(output_size=(14, 14), spatial_scale=0.03125, sampling_ratio=0, aligned=True)
      )
    )
    (mask_head): MaskRCNNConvUpsampleHead(
      (mask_fcn1): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (mask_fcn2): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (mask_fcn3): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (mask_fcn4): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (deconv): ConvTranspose2d(256, 256, kernel_size=(2, 2), stride=(2, 2))
      (deconv_relu): ReLU()
      (predictor): Conv2d(256, 7, kernel_size=(1, 1), stride=(1, 1))
    )
  )
)
[05/08 23:44:11 d2.data.datasets.coco]: Loaded 1951 images in COCO format from ./data/instances_vcoco_actions_train.json
[05/08 23:44:11 d2.data.build]: Removed 0 images with no usable annotations. 1951 images left.
[05/08 23:44:12 d2.data.build]: Distribution of instances among all 7 categories:
|   category    | #instances   |  category  | #instances   |  category  | #instances   |
|:-------------:|:-------------|:----------:|:-------------|:----------:|:-------------|
|     hold      | 1838         |   carry    | 236          |   point    | 15           |
|      eat      | 301          |   drink    | 63           |   stand    | 2150         |
| talk_on_phone | 167          |            |              |            |              |
|     total     | 4770         |            |              |            |              |
[05/08 23:44:12 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in training: [ResizeShortestEdge(short_edge_length=(640, 672, 704, 736, 768, 800), max_size=1333, sample_style='choice'), RandomFlip()]
[05/08 23:44:12 d2.data.build]: Using training sampler TrainingSampler
[05/08 23:44:12 d2.data.common]: Serializing 1951 elements to byte tensors and concatenating them all ...
[05/08 23:44:12 d2.data.common]: Serialized dataset takes 4.90 MiB
WARNING [05/08 23:44:12 d2.solver.build]: SOLVER.STEPS contains values larger than SOLVER.MAX_ITER. These values will be ignored.

model_final_f10217.pkl: 178MB [00:12, 14.6MB/s]                           
Skip loading parameter 'roi_heads.box_predictor.cls_score.weight' to the model due to incompatible shapes: (81, 1024) in the checkpoint but (8, 1024) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.cls_score.bias' to the model due to incompatible shapes: (81,) in the checkpoint but (8,) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.bbox_pred.weight' to the model due to incompatible shapes: (320, 1024) in the checkpoint but (28, 1024) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.bbox_pred.bias' to the model due to incompatible shapes: (320,) in the checkpoint but (28,) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.mask_head.predictor.weight' to the model due to incompatible shapes: (80, 256, 1, 1) in the checkpoint but (7, 256, 1, 1) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.mask_head.predictor.bias' to the model due to incompatible shapes: (80,) in the checkpoint but (7,) in the model! You might want to double check if this is expected.
Some model parameters or buffers are not found in the checkpoint:
roi_heads.box_predictor.bbox_pred.{bias, weight}
roi_heads.box_predictor.cls_score.{bias, weight}
roi_heads.mask_head.predictor.{bias, weight}

[05/08 23:44:30 d2.engine.train_loop]: Starting training from iteration 0

/usr/local/lib/python3.7/dist-packages/torch/functional.py:568: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at  ../aten/src/ATen/native/TensorShape.cpp:2228.)
  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]

[05/08 23:44:45 d2.utils.events]:  eta: 0:11:46  iter: 19  total_loss: 3.509  loss_cls: 2.275  loss_box_reg: 0.5488  loss_mask: 0.6903  loss_rpn_cls: 0.03365  loss_rpn_loc: 0.004783  time: 0.6797  data_time: 0.4443  lr: 3.9962e-05  max_mem: 2545M
[05/08 23:44:59 d2.utils.events]:  eta: 0:11:25  iter: 39  total_loss: 2.852  loss_cls: 1.574  loss_box_reg: 0.5157  loss_mask: 0.6691  loss_rpn_cls: 0.02743  loss_rpn_loc: 0.004821  time: 0.6894  data_time: 0.4270  lr: 7.9922e-05  max_mem: 2545M
[05/08 23:45:14 d2.utils.events]:  eta: 0:11:12  iter: 59  total_loss: 1.92  loss_cls: 0.7083  loss_box_reg: 0.5131  loss_mask: 0.6261  loss_rpn_cls: 0.01519  loss_rpn_loc: 0.005434  time: 0.7122  data_time: 0.4793  lr: 0.00011988  max_mem: 2545M
[05/08 23:45:28 d2.utils.events]:  eta: 0:10:55  iter: 79  total_loss: 1.858  loss_cls: 0.5534  loss_box_reg: 0.6597  loss_mask: 0.5698  loss_rpn_cls: 0.01562  loss_rpn_loc: 0.00593  time: 0.7083  data_time: 0.4169  lr: 0.00015984  max_mem: 2545M
[05/08 23:45:43 d2.utils.events]:  eta: 0:10:41  iter: 99  total_loss: 1.649  loss_cls: 0.4709  loss_box_reg: 0.629  loss_mask: 0.5024  loss_rpn_cls: 0.01413  loss_rpn_loc: 0.003994  time: 0.7145  data_time: 0.4705  lr: 0.0001998  max_mem: 2758M
[05/08 23:45:58 d2.utils.events]:  eta: 0:10:26  iter: 119  total_loss: 1.417  loss_cls: 0.3935  loss_box_reg: 0.5969  loss_mask: 0.4463  loss_rpn_cls: 0.01225  loss_rpn_loc: 0.004268  time: 0.7216  data_time: 0.4856  lr: 0.00023976  max_mem: 2758M
[05/08 23:46:12 d2.utils.events]:  eta: 0:10:11  iter: 139  total_loss: 1.177  loss_cls: 0.3305  loss_box_reg: 0.5064  loss_mask: 0.329  loss_rpn_cls: 0.009698  loss_rpn_loc: 0.005238  time: 0.7206  data_time: 0.4357  lr: 0.00027972  max_mem: 2758M
[05/08 23:46:25 d2.utils.events]:  eta: 0:09:51  iter: 159  total_loss: 0.9623  loss_cls: 0.2355  loss_box_reg: 0.4394  loss_mask: 0.2282  loss_rpn_cls: 0.007012  loss_rpn_loc: 0.004906  time: 0.7102  data_time: 0.3691  lr: 0.00031968  max_mem: 2758M
[05/08 23:46:39 d2.utils.events]:  eta: 0:09:34  iter: 179  total_loss: 0.9734  loss_cls: 0.2494  loss_box_reg: 0.4906  loss_mask: 0.2062  loss_rpn_cls: 0.007104  loss_rpn_loc: 0.00683  time: 0.7070  data_time: 0.4068  lr: 0.00035964  max_mem: 2758M
[05/08 23:46:52 d2.utils.events]:  eta: 0:09:17  iter: 199  total_loss: 0.7626  loss_cls: 0.2272  loss_box_reg: 0.3877  loss_mask: 0.1662  loss_rpn_cls: 0.004023  loss_rpn_loc: 0.005579  time: 0.7018  data_time: 0.3963  lr: 0.0003996  max_mem: 2758M
[05/08 23:47:06 d2.utils.events]:  eta: 0:08:59  iter: 219  total_loss: 0.6499  loss_cls: 0.1997  loss_box_reg: 0.2687  loss_mask: 0.1704  loss_rpn_cls: 0.006145  loss_rpn_loc: 0.01117  time: 0.7044  data_time: 0.4561  lr: 0.00043956  max_mem: 2758M
[05/08 23:47:20 d2.utils.events]:  eta: 0:08:45  iter: 239  total_loss: 0.6414  loss_cls: 0.1949  loss_box_reg: 0.2449  loss_mask: 0.1675  loss_rpn_cls: 0.007746  loss_rpn_loc: 0.006541  time: 0.7012  data_time: 0.3887  lr: 0.00047952  max_mem: 2758M
[05/08 23:47:34 d2.utils.events]:  eta: 0:08:32  iter: 259  total_loss: 0.6398  loss_cls: 0.1775  loss_box_reg: 0.2551  loss_mask: 0.1514  loss_rpn_cls: 0.003853  loss_rpn_loc: 0.007165  time: 0.7021  data_time: 0.4405  lr: 0.00051948  max_mem: 2758M
[05/08 23:47:48 d2.utils.events]:  eta: 0:08:17  iter: 279  total_loss: 0.6898  loss_cls: 0.1958  loss_box_reg: 0.2795  loss_mask: 0.1758  loss_rpn_cls: 0.002961  loss_rpn_loc: 0.00508  time: 0.7004  data_time: 0.4070  lr: 0.00055944  max_mem: 2758M
[05/08 23:48:01 d2.utils.events]:  eta: 0:08:02  iter: 299  total_loss: 0.6019  loss_cls: 0.2215  loss_box_reg: 0.2301  loss_mask: 0.1476  loss_rpn_cls: 0.001727  loss_rpn_loc: 0.006422  time: 0.6976  data_time: 0.3794  lr: 0.0005994  max_mem: 2758M
[05/08 23:48:13 d2.utils.events]:  eta: 0:07:48  iter: 319  total_loss: 0.5421  loss_cls: 0.1544  loss_box_reg: 0.2133  loss_mask: 0.1664  loss_rpn_cls: 0.004455  loss_rpn_loc: 0.005853  time: 0.6936  data_time: 0.3608  lr: 0.00063936  max_mem: 2758M
[05/08 23:48:27 d2.utils.events]:  eta: 0:07:32  iter: 339  total_loss: 0.6201  loss_cls: 0.1972  loss_box_reg: 0.2262  loss_mask: 0.1595  loss_rpn_cls: 0.001733  loss_rpn_loc: 0.005763  time: 0.6922  data_time: 0.3985  lr: 0.00067932  max_mem: 2758M
[05/08 23:48:40 d2.utils.events]:  eta: 0:07:17  iter: 359  total_loss: 0.5356  loss_cls: 0.1841  loss_box_reg: 0.1714  loss_mask: 0.1613  loss_rpn_cls: 0.001398  loss_rpn_loc: 0.003595  time: 0.6914  data_time: 0.4123  lr: 0.00071928  max_mem: 2758M
[05/08 23:48:55 d2.utils.events]:  eta: 0:07:03  iter: 379  total_loss: 0.5604  loss_cls: 0.1602  loss_box_reg: 0.2137  loss_mask: 0.1598  loss_rpn_cls: 0.003006  loss_rpn_loc: 0.007538  time: 0.6920  data_time: 0.4284  lr: 0.00075924  max_mem: 2758M
[05/08 23:49:08 d2.utils.events]:  eta: 0:06:50  iter: 399  total_loss: 0.5451  loss_cls: 0.1769  loss_box_reg: 0.1712  loss_mask: 0.152  loss_rpn_cls: 0.0008314  loss_rpn_loc: 0.005317  time: 0.6920  data_time: 0.4265  lr: 0.0007992  max_mem: 2758M
[05/08 23:49:23 d2.utils.events]:  eta: 0:06:37  iter: 419  total_loss: 0.6469  loss_cls: 0.1963  loss_box_reg: 0.2509  loss_mask: 0.1429  loss_rpn_cls: 0.001343  loss_rpn_loc: 0.004646  time: 0.6942  data_time: 0.4671  lr: 0.00083916  max_mem: 2758M
[05/08 23:49:38 d2.utils.events]:  eta: 0:06:22  iter: 439  total_loss: 0.6079  loss_cls: 0.1721  loss_box_reg: 0.2213  loss_mask: 0.1614  loss_rpn_cls: 0.002106  loss_rpn_loc: 0.006757  time: 0.6952  data_time: 0.4372  lr: 0.00087912  max_mem: 2758M
[05/08 23:49:53 d2.utils.events]:  eta: 0:06:09  iter: 459  total_loss: 0.5329  loss_cls: 0.1746  loss_box_reg: 0.1964  loss_mask: 0.1625  loss_rpn_cls: 0.0007963  loss_rpn_loc: 0.005657  time: 0.6980  data_time: 0.4804  lr: 0.00091908  max_mem: 2758M
[05/08 23:50:06 d2.utils.events]:  eta: 0:05:55  iter: 479  total_loss: 0.6376  loss_cls: 0.2231  loss_box_reg: 0.2272  loss_mask: 0.1416  loss_rpn_cls: 0.000534  loss_rpn_loc: 0.005627  time: 0.6962  data_time: 0.3852  lr: 0.00095904  max_mem: 2758M
[05/08 23:50:19 d2.utils.events]:  eta: 0:05:41  iter: 499  total_loss: 0.5207  loss_cls: 0.1565  loss_box_reg: 0.2171  loss_mask: 0.1651  loss_rpn_cls: 0.001735  loss_rpn_loc: 0.008417  time: 0.6955  data_time: 0.3949  lr: 0.000999  max_mem: 2758M
[05/08 23:50:32 d2.utils.events]:  eta: 0:05:27  iter: 519  total_loss: 0.6803  loss_cls: 0.1874  loss_box_reg: 0.2232  loss_mask: 0.211  loss_rpn_cls: 0.001165  loss_rpn_loc: 0.006802  time: 0.6938  data_time: 0.3818  lr: 0.001039  max_mem: 2758M
[05/08 23:50:46 d2.utils.events]:  eta: 0:05:13  iter: 539  total_loss: 0.6538  loss_cls: 0.2187  loss_box_reg: 0.2295  loss_mask: 0.1532  loss_rpn_cls: 0.001852  loss_rpn_loc: 0.005097  time: 0.6929  data_time: 0.3954  lr: 0.0010789  max_mem: 2758M
[05/08 23:50:59 d2.utils.events]:  eta: 0:04:59  iter: 559  total_loss: 0.607  loss_cls: 0.179  loss_box_reg: 0.1992  loss_mask: 0.1563  loss_rpn_cls: 0.00148  loss_rpn_loc: 0.004432  time: 0.6922  data_time: 0.3964  lr: 0.0011189  max_mem: 2758M
[05/08 23:51:14 d2.utils.events]:  eta: 0:04:46  iter: 579  total_loss: 0.6847  loss_cls: 0.2268  loss_box_reg: 0.2463  loss_mask: 0.1967  loss_rpn_cls: 0.002075  loss_rpn_loc: 0.006982  time: 0.6931  data_time: 0.4470  lr: 0.0011588  max_mem: 2758M
[05/08 23:51:27 d2.utils.events]:  eta: 0:04:32  iter: 599  total_loss: 0.6654  loss_cls: 0.2124  loss_box_reg: 0.264  loss_mask: 0.1868  loss_rpn_cls: 0.002221  loss_rpn_loc: 0.006894  time: 0.6924  data_time: 0.3940  lr: 0.0011988  max_mem: 2758M
[05/08 23:51:40 d2.utils.events]:  eta: 0:04:18  iter: 619  total_loss: 0.5272  loss_cls: 0.1701  loss_box_reg: 0.1946  loss_mask: 0.1323  loss_rpn_cls: 0.001854  loss_rpn_loc: 0.003515  time: 0.6912  data_time: 0.3696  lr: 0.0012388  max_mem: 2758M
[05/08 23:51:54 d2.utils.events]:  eta: 0:04:04  iter: 639  total_loss: 0.586  loss_cls: 0.1475  loss_box_reg: 0.2384  loss_mask: 0.1607  loss_rpn_cls: 0.0008062  loss_rpn_loc: 0.005421  time: 0.6906  data_time: 0.4001  lr: 0.0012787  max_mem: 2758M
[05/08 23:52:07 d2.utils.events]:  eta: 0:03:51  iter: 659  total_loss: 0.5878  loss_cls: 0.1659  loss_box_reg: 0.2227  loss_mask: 0.1995  loss_rpn_cls: 0.003757  loss_rpn_loc: 0.004428  time: 0.6896  data_time: 0.3835  lr: 0.0013187  max_mem: 2758M
[05/08 23:52:20 d2.utils.events]:  eta: 0:03:37  iter: 679  total_loss: 0.6112  loss_cls: 0.1709  loss_box_reg: 0.2152  loss_mask: 0.1502  loss_rpn_cls: 0.000707  loss_rpn_loc: 0.005767  time: 0.6881  data_time: 0.3700  lr: 0.0013586  max_mem: 2758M
[05/08 23:52:34 d2.utils.events]:  eta: 0:03:23  iter: 699  total_loss: 0.567  loss_cls: 0.2016  loss_box_reg: 0.1862  loss_mask: 0.1611  loss_rpn_cls: 0.00108  loss_rpn_loc: 0.005898  time: 0.6886  data_time: 0.4387  lr: 0.0013986  max_mem: 2758M
[05/08 23:52:48 d2.utils.events]:  eta: 0:03:09  iter: 719  total_loss: 0.5998  loss_cls: 0.1429  loss_box_reg: 0.2213  loss_mask: 0.1731  loss_rpn_cls: 0.001619  loss_rpn_loc: 0.007896  time: 0.6886  data_time: 0.4136  lr: 0.0014386  max_mem: 2758M
[05/08 23:53:01 d2.utils.events]:  eta: 0:02:55  iter: 739  total_loss: 0.7415  loss_cls: 0.2587  loss_box_reg: 0.259  loss_mask: 0.2077  loss_rpn_cls: 0.003711  loss_rpn_loc: 0.004918  time: 0.6874  data_time: 0.3692  lr: 0.0014785  max_mem: 2758M
[05/08 23:53:14 d2.utils.events]:  eta: 0:02:42  iter: 759  total_loss: 0.5897  loss_cls: 0.1851  loss_box_reg: 0.2103  loss_mask: 0.1915  loss_rpn_cls: 0.001144  loss_rpn_loc: 0.00487  time: 0.6873  data_time: 0.4155  lr: 0.0015185  max_mem: 2758M
[05/08 23:53:27 d2.utils.events]:  eta: 0:02:28  iter: 779  total_loss: 0.6183  loss_cls: 0.1405  loss_box_reg: 0.2124  loss_mask: 0.1609  loss_rpn_cls: 0.002169  loss_rpn_loc: 0.006691  time: 0.6866  data_time: 0.3883  lr: 0.0015584  max_mem: 2758M
[05/08 23:53:41 d2.utils.events]:  eta: 0:02:15  iter: 799  total_loss: 0.6244  loss_cls: 0.1784  loss_box_reg: 0.2373  loss_mask: 0.1556  loss_rpn_cls: 0.001868  loss_rpn_loc: 0.00604  time: 0.6860  data_time: 0.3901  lr: 0.0015984  max_mem: 2758M
[05/08 23:53:54 d2.utils.events]:  eta: 0:02:01  iter: 819  total_loss: 0.532  loss_cls: 0.1634  loss_box_reg: 0.1875  loss_mask: 0.1526  loss_rpn_cls: 0.001021  loss_rpn_loc: 0.004913  time: 0.6850  data_time: 0.3742  lr: 0.0016384  max_mem: 2758M
[05/08 23:54:07 d2.utils.events]:  eta: 0:01:47  iter: 839  total_loss: 0.5917  loss_cls: 0.1534  loss_box_reg: 0.2076  loss_mask: 0.1468  loss_rpn_cls: 0.001153  loss_rpn_loc: 0.006929  time: 0.6842  data_time: 0.3797  lr: 0.0016783  max_mem: 2758M
[05/08 23:54:19 d2.utils.events]:  eta: 0:01:34  iter: 859  total_loss: 0.6272  loss_cls: 0.2112  loss_box_reg: 0.2271  loss_mask: 0.1577  loss_rpn_cls: 0.002351  loss_rpn_loc: 0.006446  time: 0.6824  data_time: 0.3372  lr: 0.0017183  max_mem: 2758M
[05/08 23:54:32 d2.utils.events]:  eta: 0:01:20  iter: 879  total_loss: 0.5061  loss_cls: 0.1737  loss_box_reg: 0.1993  loss_mask: 0.1525  loss_rpn_cls: 0.001229  loss_rpn_loc: 0.005237  time: 0.6817  data_time: 0.3815  lr: 0.0017582  max_mem: 2758M
[05/08 23:54:46 d2.utils.events]:  eta: 0:01:07  iter: 899  total_loss: 0.6134  loss_cls: 0.1872  loss_box_reg: 0.2271  loss_mask: 0.156  loss_rpn_cls: 0.0004449  loss_rpn_loc: 0.006331  time: 0.6822  data_time: 0.4234  lr: 0.0017982  max_mem: 2758M
[05/08 23:55:00 d2.utils.events]:  eta: 0:00:53  iter: 919  total_loss: 0.6265  loss_cls: 0.2008  loss_box_reg: 0.2045  loss_mask: 0.1336  loss_rpn_cls: 0.001015  loss_rpn_loc: 0.005531  time: 0.6823  data_time: 0.4159  lr: 0.0018382  max_mem: 2758M
[05/08 23:55:13 d2.utils.events]:  eta: 0:00:40  iter: 939  total_loss: 0.5976  loss_cls: 0.2227  loss_box_reg: 0.1932  loss_mask: 0.1524  loss_rpn_cls: 0.0007851  loss_rpn_loc: 0.00524  time: 0.6823  data_time: 0.4041  lr: 0.0018781  max_mem: 2758M
[05/08 23:55:27 d2.utils.events]:  eta: 0:00:26  iter: 959  total_loss: 0.5818  loss_cls: 0.1942  loss_box_reg: 0.1894  loss_mask: 0.1421  loss_rpn_cls: 0.001578  loss_rpn_loc: 0.005245  time: 0.6822  data_time: 0.3996  lr: 0.0019181  max_mem: 2758M
[05/08 23:55:38 d2.utils.events]:  eta: 0:00:13  iter: 979  total_loss: 0.6051  loss_cls: 0.2029  loss_box_reg: 0.2244  loss_mask: 0.1565  loss_rpn_cls: 0.001146  loss_rpn_loc: 0.006171  time: 0.6794  data_time: 0.2656  lr: 0.001958  max_mem: 2758M
[05/08 23:55:45 d2.utils.events]:  eta: 0:00:00  iter: 999  total_loss: 0.5994  loss_cls: 0.17  loss_box_reg: 0.2439  loss_mask: 0.1669  loss_rpn_cls: 0.001093  loss_rpn_loc: 0.006357  time: 0.6719  data_time: 0.0140  lr: 0.001998  max_mem: 2758M
[05/08 23:55:45 d2.engine.hooks]: Overall training speed: 998 iterations in 0:11:10 (0.6719 s / it)
[05/08 23:55:45 d2.engine.hooks]: Total training time: 0:11:12 (0:00:02 on hooks)

Now, we perform inference with the trained model on the fruits_nuts dataset. First, let's create a predictor using the model we just trained:

Run Eval Metrics for 0.5

cfg.MODEL.WEIGHTS = os.path.join(cfg.OUTPUT_DIR, "model_final.pth")
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5   # set the testing threshold for this model
cfg.DATASETS.TEST = ("rel_actions_val3", )
predictor = DefaultPredictor(cfg)

[05/09 00:15:16 d2.checkpoint.c2_model_loading]: Following weights matched with model:
| Names in Model                                  | Names in Checkpoint                                                                                  | Shapes                                          |
|:------------------------------------------------|:-----------------------------------------------------------------------------------------------------|:------------------------------------------------|
| backbone.bottom_up.res2.0.conv1.*               | backbone.bottom_up.res2.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,1,1)             |
| backbone.bottom_up.res2.0.conv2.*               | backbone.bottom_up.res2.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.0.conv3.*               | backbone.bottom_up.res2.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.0.shortcut.*            | backbone.bottom_up.res2.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.1.conv1.*               | backbone.bottom_up.res2.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.1.conv2.*               | backbone.bottom_up.res2.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.1.conv3.*               | backbone.bottom_up.res2.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.2.conv1.*               | backbone.bottom_up.res2.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.2.conv2.*               | backbone.bottom_up.res2.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.2.conv3.*               | backbone.bottom_up.res2.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res3.0.conv1.*               | backbone.bottom_up.res3.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,256,1,1)       |
| backbone.bottom_up.res3.0.conv2.*               | backbone.bottom_up.res3.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.0.conv3.*               | backbone.bottom_up.res3.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.0.shortcut.*            | backbone.bottom_up.res3.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (512,) (512,) (512,) (512,) (512,256,1,1)       |
| backbone.bottom_up.res3.1.conv1.*               | backbone.bottom_up.res3.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.1.conv2.*               | backbone.bottom_up.res3.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.1.conv3.*               | backbone.bottom_up.res3.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.2.conv1.*               | backbone.bottom_up.res3.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.2.conv2.*               | backbone.bottom_up.res3.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.2.conv3.*               | backbone.bottom_up.res3.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.3.conv1.*               | backbone.bottom_up.res3.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.3.conv2.*               | backbone.bottom_up.res3.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.3.conv3.*               | backbone.bottom_up.res3.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res4.0.conv1.*               | backbone.bottom_up.res4.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,512,1,1)       |
| backbone.bottom_up.res4.0.conv2.*               | backbone.bottom_up.res4.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.0.conv3.*               | backbone.bottom_up.res4.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.0.shortcut.*            | backbone.bottom_up.res4.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (1024,) (1024,) (1024,) (1024,) (1024,512,1,1)  |
| backbone.bottom_up.res4.1.conv1.*               | backbone.bottom_up.res4.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.1.conv2.*               | backbone.bottom_up.res4.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.1.conv3.*               | backbone.bottom_up.res4.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.2.conv1.*               | backbone.bottom_up.res4.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.2.conv2.*               | backbone.bottom_up.res4.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.2.conv3.*               | backbone.bottom_up.res4.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.3.conv1.*               | backbone.bottom_up.res4.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.3.conv2.*               | backbone.bottom_up.res4.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.3.conv3.*               | backbone.bottom_up.res4.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.4.conv1.*               | backbone.bottom_up.res4.4.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.4.conv2.*               | backbone.bottom_up.res4.4.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.4.conv3.*               | backbone.bottom_up.res4.4.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.5.conv1.*               | backbone.bottom_up.res4.5.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.5.conv2.*               | backbone.bottom_up.res4.5.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.5.conv3.*               | backbone.bottom_up.res4.5.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res5.0.conv1.*               | backbone.bottom_up.res5.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,1024,1,1)      |
| backbone.bottom_up.res5.0.conv2.*               | backbone.bottom_up.res5.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.0.conv3.*               | backbone.bottom_up.res5.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.0.shortcut.*            | backbone.bottom_up.res5.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (2048,) (2048,) (2048,) (2048,) (2048,1024,1,1) |
| backbone.bottom_up.res5.1.conv1.*               | backbone.bottom_up.res5.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.1.conv2.*               | backbone.bottom_up.res5.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.1.conv3.*               | backbone.bottom_up.res5.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.2.conv1.*               | backbone.bottom_up.res5.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.2.conv2.*               | backbone.bottom_up.res5.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.2.conv3.*               | backbone.bottom_up.res5.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.stem.conv1.*                 | backbone.bottom_up.stem.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}      | (64,) (64,) (64,) (64,) (64,3,7,7)              |
| backbone.fpn_lateral2.*                         | backbone.fpn_lateral2.{bias,weight}                                                                  | (256,) (256,256,1,1)                            |
| backbone.fpn_lateral3.*                         | backbone.fpn_lateral3.{bias,weight}                                                                  | (256,) (256,512,1,1)                            |
| backbone.fpn_lateral4.*                         | backbone.fpn_lateral4.{bias,weight}                                                                  | (256,) (256,1024,1,1)                           |
| backbone.fpn_lateral5.*                         | backbone.fpn_lateral5.{bias,weight}                                                                  | (256,) (256,2048,1,1)                           |
| backbone.fpn_output2.*                          | backbone.fpn_output2.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output3.*                          | backbone.fpn_output3.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output4.*                          | backbone.fpn_output4.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output5.*                          | backbone.fpn_output5.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.anchor_deltas.*     | proposal_generator.rpn_head.anchor_deltas.{bias,weight}                                              | (12,) (12,256,1,1)                              |
| proposal_generator.rpn_head.conv.*              | proposal_generator.rpn_head.conv.{bias,weight}                                                       | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.objectness_logits.* | proposal_generator.rpn_head.objectness_logits.{bias,weight}                                          | (3,) (3,256,1,1)                                |
| roi_heads.box_head.fc1.*                        | roi_heads.box_head.fc1.{bias,weight}                                                                 | (1024,) (1024,12544)                            |
| roi_heads.box_head.fc2.*                        | roi_heads.box_head.fc2.{bias,weight}                                                                 | (1024,) (1024,1024)                             |
| roi_heads.box_predictor.bbox_pred.*             | roi_heads.box_predictor.bbox_pred.{bias,weight}                                                      | (28,) (28,1024)                                 |
| roi_heads.box_predictor.cls_score.*             | roi_heads.box_predictor.cls_score.{bias,weight}                                                      | (8,) (8,1024)                                   |
| roi_heads.mask_head.deconv.*                    | roi_heads.mask_head.deconv.{bias,weight}                                                             | (256,) (256,256,2,2)                            |
| roi_heads.mask_head.mask_fcn1.*                 | roi_heads.mask_head.mask_fcn1.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn2.*                 | roi_heads.mask_head.mask_fcn2.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn3.*                 | roi_heads.mask_head.mask_fcn3.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn4.*                 | roi_heads.mask_head.mask_fcn4.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.predictor.*                 | roi_heads.mask_head.predictor.{bias,weight}                                                          | (7,) (7,256,1,1)                                |

Then, we randomly select several samples to visualize the prediction results.

from detectron2.utils.visualizer import ColorMode

for d in random.sample(train_dicts, 3):    
    im = cv2.imread(d["file_name"])
    outputs = predictor(im)
    v = Visualizer(im[:, :, ::-1],
                   metadata=train_metadata, 
                   scale=0.8, 
                   instance_mode=ColorMode.IMAGE_BW   # remove the colors of unsegmented pixels
    )
    v = v.draw_instance_predictions(outputs["instances"].to("cpu"))
    cv2_imshow(v.get_image()[:, :, ::-1])

from detectron2.evaluation import COCOEvaluator, inference_on_dataset
from detectron2.data import build_detection_test_loader
from detectron2.engine import DefaultTrainer

evaluator = COCOEvaluator("rel_actions_val3", cfg, False, output_dir="./output_evals/")
val_loader = build_detection_test_loader(cfg, "rel_actions_val3")
inference_on_dataset(trainer.model, val_loader, evaluator)

WARNING [05/09 00:15:25 d2.evaluation.coco_evaluation]: COCO Evaluator instantiated using config, this is deprecated behavior. Please pass in explicit arguments instead.
[05/09 00:15:25 d2.data.datasets.coco]: Loaded 2215 images in COCO format from ./data/instances_vcoco_actions_val_2014.json
[05/09 00:15:26 d2.data.build]: Distribution of instances among all 7 categories:
|   category    | #instances   |  category  | #instances   |  category  | #instances   |
|:-------------:|:-------------|:----------:|:-------------|:----------:|:-------------|
|     hold      | 2163         |   carry    | 262          |   point    | 23           |
|      eat      | 376          |   drink    | 70           |   stand    | 2448         |
| talk_on_phone | 187          |            |              |            |              |
|     total     | 5529         |            |              |            |              |
[05/09 00:15:26 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in inference: [ResizeShortestEdge(short_edge_length=(800, 800), max_size=1333, sample_style='choice')]
[05/09 00:15:26 d2.data.common]: Serializing 2215 elements to byte tensors and concatenating them all ...
[05/09 00:15:26 d2.data.common]: Serialized dataset takes 5.66 MiB
[05/09 00:15:26 d2.evaluation.evaluator]: Start inference on 2215 batches
[05/09 00:15:30 d2.evaluation.evaluator]: Inference done 11/2215. Dataloading: 0.1800 s/iter. Inference: 0.0651 s/iter. Eval: 0.0063 s/iter. Total: 0.2513 s/iter. ETA=0:09:13
[05/09 00:15:35 d2.evaluation.evaluator]: Inference done 28/2215. Dataloading: 0.2276 s/iter. Inference: 0.0669 s/iter. Eval: 0.0073 s/iter. Total: 0.3020 s/iter. ETA=0:11:00
[05/09 00:15:41 d2.evaluation.evaluator]: Inference done 44/2215. Dataloading: 0.2372 s/iter. Inference: 0.0679 s/iter. Eval: 0.0079 s/iter. Total: 0.3132 s/iter. ETA=0:11:19
[05/09 00:15:46 d2.evaluation.evaluator]: Inference done 58/2215. Dataloading: 0.2552 s/iter. Inference: 0.0675 s/iter. Eval: 0.0081 s/iter. Total: 0.3310 s/iter. ETA=0:11:53
[05/09 00:15:51 d2.evaluation.evaluator]: Inference done 73/2215. Dataloading: 0.2561 s/iter. Inference: 0.0684 s/iter. Eval: 0.0080 s/iter. Total: 0.3326 s/iter. ETA=0:11:52
[05/09 00:15:56 d2.evaluation.evaluator]: Inference done 88/2215. Dataloading: 0.2577 s/iter. Inference: 0.0684 s/iter. Eval: 0.0081 s/iter. Total: 0.3345 s/iter. ETA=0:11:51
[05/09 00:16:02 d2.evaluation.evaluator]: Inference done 104/2215. Dataloading: 0.2600 s/iter. Inference: 0.0681 s/iter. Eval: 0.0088 s/iter. Total: 0.3371 s/iter. ETA=0:11:51
[05/09 00:16:07 d2.evaluation.evaluator]: Inference done 120/2215. Dataloading: 0.2593 s/iter. Inference: 0.0681 s/iter. Eval: 0.0088 s/iter. Total: 0.3363 s/iter. ETA=0:11:44
[05/09 00:16:13 d2.evaluation.evaluator]: Inference done 136/2215. Dataloading: 0.2606 s/iter. Inference: 0.0679 s/iter. Eval: 0.0087 s/iter. Total: 0.3374 s/iter. ETA=0:11:41
[05/09 00:16:18 d2.evaluation.evaluator]: Inference done 152/2215. Dataloading: 0.2592 s/iter. Inference: 0.0680 s/iter. Eval: 0.0089 s/iter. Total: 0.3363 s/iter. ETA=0:11:33
[05/09 00:16:23 d2.evaluation.evaluator]: Inference done 168/2215. Dataloading: 0.2594 s/iter. Inference: 0.0679 s/iter. Eval: 0.0086 s/iter. Total: 0.3360 s/iter. ETA=0:11:27
[05/09 00:16:29 d2.evaluation.evaluator]: Inference done 184/2215. Dataloading: 0.2593 s/iter. Inference: 0.0680 s/iter. Eval: 0.0087 s/iter. Total: 0.3362 s/iter. ETA=0:11:22
[05/09 00:16:34 d2.evaluation.evaluator]: Inference done 200/2215. Dataloading: 0.2593 s/iter. Inference: 0.0680 s/iter. Eval: 0.0087 s/iter. Total: 0.3363 s/iter. ETA=0:11:17
[05/09 00:16:39 d2.evaluation.evaluator]: Inference done 216/2215. Dataloading: 0.2586 s/iter. Inference: 0.0681 s/iter. Eval: 0.0088 s/iter. Total: 0.3357 s/iter. ETA=0:11:11
[05/09 00:16:44 d2.evaluation.evaluator]: Inference done 231/2215. Dataloading: 0.2597 s/iter. Inference: 0.0682 s/iter. Eval: 0.0087 s/iter. Total: 0.3368 s/iter. ETA=0:11:08
[05/09 00:16:50 d2.evaluation.evaluator]: Inference done 246/2215. Dataloading: 0.2601 s/iter. Inference: 0.0681 s/iter. Eval: 0.0086 s/iter. Total: 0.3370 s/iter. ETA=0:11:03
[05/09 00:16:55 d2.evaluation.evaluator]: Inference done 262/2215. Dataloading: 0.2603 s/iter. Inference: 0.0682 s/iter. Eval: 0.0085 s/iter. Total: 0.3371 s/iter. ETA=0:10:58
[05/09 00:17:00 d2.evaluation.evaluator]: Inference done 276/2215. Dataloading: 0.2617 s/iter. Inference: 0.0682 s/iter. Eval: 0.0085 s/iter. Total: 0.3387 s/iter. ETA=0:10:56
[05/09 00:17:05 d2.evaluation.evaluator]: Inference done 289/2215. Dataloading: 0.2638 s/iter. Inference: 0.0683 s/iter. Eval: 0.0086 s/iter. Total: 0.3409 s/iter. ETA=0:10:56
[05/09 00:17:10 d2.evaluation.evaluator]: Inference done 305/2215. Dataloading: 0.2627 s/iter. Inference: 0.0682 s/iter. Eval: 0.0086 s/iter. Total: 0.3397 s/iter. ETA=0:10:48
[05/09 00:17:15 d2.evaluation.evaluator]: Inference done 319/2215. Dataloading: 0.2638 s/iter. Inference: 0.0682 s/iter. Eval: 0.0085 s/iter. Total: 0.3407 s/iter. ETA=0:10:45
[05/09 00:17:21 d2.evaluation.evaluator]: Inference done 334/2215. Dataloading: 0.2653 s/iter. Inference: 0.0681 s/iter. Eval: 0.0084 s/iter. Total: 0.3420 s/iter. ETA=0:10:43
[05/09 00:17:26 d2.evaluation.evaluator]: Inference done 348/2215. Dataloading: 0.2671 s/iter. Inference: 0.0681 s/iter. Eval: 0.0083 s/iter. Total: 0.3438 s/iter. ETA=0:10:41
[05/09 00:17:50 d2.evaluation.evaluator]: Inference done 359/2215. Dataloading: 0.2652 s/iter. Inference: 0.1269 s/iter. Eval: 0.0084 s/iter. Total: 0.4007 s/iter. ETA=0:12:23
[05/09 00:17:55 d2.evaluation.evaluator]: Inference done 379/2215. Dataloading: 0.2603 s/iter. Inference: 0.1238 s/iter. Eval: 0.0084 s/iter. Total: 0.3927 s/iter. ETA=0:12:00
[05/09 00:18:01 d2.evaluation.evaluator]: Inference done 394/2215. Dataloading: 0.2616 s/iter. Inference: 0.1216 s/iter. Eval: 0.0083 s/iter. Total: 0.3917 s/iter. ETA=0:11:53
[05/09 00:18:06 d2.evaluation.evaluator]: Inference done 410/2215. Dataloading: 0.2620 s/iter. Inference: 0.1195 s/iter. Eval: 0.0083 s/iter. Total: 0.3900 s/iter. ETA=0:11:43
[05/09 00:18:12 d2.evaluation.evaluator]: Inference done 425/2215. Dataloading: 0.2632 s/iter. Inference: 0.1177 s/iter. Eval: 0.0082 s/iter. Total: 0.3893 s/iter. ETA=0:11:36
[05/09 00:18:17 d2.evaluation.evaluator]: Inference done 440/2215. Dataloading: 0.2637 s/iter. Inference: 0.1160 s/iter. Eval: 0.0082 s/iter. Total: 0.3881 s/iter. ETA=0:11:28
[05/09 00:18:23 d2.evaluation.evaluator]: Inference done 456/2215. Dataloading: 0.2634 s/iter. Inference: 0.1143 s/iter. Eval: 0.0083 s/iter. Total: 0.3862 s/iter. ETA=0:11:19
[05/09 00:18:28 d2.evaluation.evaluator]: Inference done 472/2215. Dataloading: 0.2626 s/iter. Inference: 0.1127 s/iter. Eval: 0.0083 s/iter. Total: 0.3838 s/iter. ETA=0:11:09
[05/09 00:18:33 d2.evaluation.evaluator]: Inference done 487/2215. Dataloading: 0.2627 s/iter. Inference: 0.1113 s/iter. Eval: 0.0083 s/iter. Total: 0.3825 s/iter. ETA=0:11:00
[05/09 00:18:38 d2.evaluation.evaluator]: Inference done 502/2215. Dataloading: 0.2631 s/iter. Inference: 0.1100 s/iter. Eval: 0.0083 s/iter. Total: 0.3816 s/iter. ETA=0:10:53
[05/09 00:18:44 d2.evaluation.evaluator]: Inference done 518/2215. Dataloading: 0.2638 s/iter. Inference: 0.1087 s/iter. Eval: 0.0083 s/iter. Total: 0.3810 s/iter. ETA=0:10:46
[05/09 00:18:49 d2.evaluation.evaluator]: Inference done 534/2215. Dataloading: 0.2638 s/iter. Inference: 0.1074 s/iter. Eval: 0.0083 s/iter. Total: 0.3798 s/iter. ETA=0:10:38
[05/09 00:18:55 d2.evaluation.evaluator]: Inference done 552/2215. Dataloading: 0.2624 s/iter. Inference: 0.1061 s/iter. Eval: 0.0083 s/iter. Total: 0.3771 s/iter. ETA=0:10:27
[05/09 00:19:00 d2.evaluation.evaluator]: Inference done 568/2215. Dataloading: 0.2619 s/iter. Inference: 0.1051 s/iter. Eval: 0.0083 s/iter. Total: 0.3755 s/iter. ETA=0:10:18
[05/09 00:19:05 d2.evaluation.evaluator]: Inference done 583/2215. Dataloading: 0.2618 s/iter. Inference: 0.1041 s/iter. Eval: 0.0083 s/iter. Total: 0.3744 s/iter. ETA=0:10:11
[05/09 00:19:10 d2.evaluation.evaluator]: Inference done 598/2215. Dataloading: 0.2619 s/iter. Inference: 0.1031 s/iter. Eval: 0.0083 s/iter. Total: 0.3736 s/iter. ETA=0:10:04
[05/09 00:19:15 d2.evaluation.evaluator]: Inference done 615/2215. Dataloading: 0.2612 s/iter. Inference: 0.1022 s/iter. Eval: 0.0083 s/iter. Total: 0.3719 s/iter. ETA=0:09:54
[05/09 00:19:21 d2.evaluation.evaluator]: Inference done 632/2215. Dataloading: 0.2606 s/iter. Inference: 0.1012 s/iter. Eval: 0.0084 s/iter. Total: 0.3703 s/iter. ETA=0:09:46
[05/09 00:19:26 d2.evaluation.evaluator]: Inference done 649/2215. Dataloading: 0.2603 s/iter. Inference: 0.1003 s/iter. Eval: 0.0083 s/iter. Total: 0.3692 s/iter. ETA=0:09:38
[05/09 00:19:32 d2.evaluation.evaluator]: Inference done 665/2215. Dataloading: 0.2607 s/iter. Inference: 0.0995 s/iter. Eval: 0.0083 s/iter. Total: 0.3688 s/iter. ETA=0:09:31
[05/09 00:19:38 d2.evaluation.evaluator]: Inference done 681/2215. Dataloading: 0.2614 s/iter. Inference: 0.0988 s/iter. Eval: 0.0084 s/iter. Total: 0.3688 s/iter. ETA=0:09:25
[05/09 00:19:43 d2.evaluation.evaluator]: Inference done 695/2215. Dataloading: 0.2618 s/iter. Inference: 0.0982 s/iter. Eval: 0.0084 s/iter. Total: 0.3685 s/iter. ETA=0:09:20
[05/09 00:19:48 d2.evaluation.evaluator]: Inference done 711/2215. Dataloading: 0.2618 s/iter. Inference: 0.0975 s/iter. Eval: 0.0084 s/iter. Total: 0.3679 s/iter. ETA=0:09:13
[05/09 00:19:53 d2.evaluation.evaluator]: Inference done 727/2215. Dataloading: 0.2615 s/iter. Inference: 0.0969 s/iter. Eval: 0.0084 s/iter. Total: 0.3670 s/iter. ETA=0:09:06
[05/09 00:19:59 d2.evaluation.evaluator]: Inference done 741/2215. Dataloading: 0.2621 s/iter. Inference: 0.0964 s/iter. Eval: 0.0085 s/iter. Total: 0.3671 s/iter. ETA=0:09:01
[05/09 00:20:04 d2.evaluation.evaluator]: Inference done 757/2215. Dataloading: 0.2622 s/iter. Inference: 0.0958 s/iter. Eval: 0.0085 s/iter. Total: 0.3667 s/iter. ETA=0:08:54
[05/09 00:20:09 d2.evaluation.evaluator]: Inference done 773/2215. Dataloading: 0.2620 s/iter. Inference: 0.0952 s/iter. Eval: 0.0085 s/iter. Total: 0.3659 s/iter. ETA=0:08:47
[05/09 00:20:15 d2.evaluation.evaluator]: Inference done 788/2215. Dataloading: 0.2621 s/iter. Inference: 0.0947 s/iter. Eval: 0.0085 s/iter. Total: 0.3655 s/iter. ETA=0:08:41
[05/09 00:20:20 d2.evaluation.evaluator]: Inference done 804/2215. Dataloading: 0.2620 s/iter. Inference: 0.0941 s/iter. Eval: 0.0085 s/iter. Total: 0.3648 s/iter. ETA=0:08:34
[05/09 00:20:25 d2.evaluation.evaluator]: Inference done 820/2215. Dataloading: 0.2617 s/iter. Inference: 0.0936 s/iter. Eval: 0.0085 s/iter. Total: 0.3640 s/iter. ETA=0:08:27
[05/09 00:20:30 d2.evaluation.evaluator]: Inference done 836/2215. Dataloading: 0.2615 s/iter. Inference: 0.0930 s/iter. Eval: 0.0085 s/iter. Total: 0.3632 s/iter. ETA=0:08:20
[05/09 00:20:35 d2.evaluation.evaluator]: Inference done 850/2215. Dataloading: 0.2620 s/iter. Inference: 0.0926 s/iter. Eval: 0.0084 s/iter. Total: 0.3633 s/iter. ETA=0:08:15
[05/09 00:20:40 d2.evaluation.evaluator]: Inference done 865/2215. Dataloading: 0.2619 s/iter. Inference: 0.0922 s/iter. Eval: 0.0085 s/iter. Total: 0.3628 s/iter. ETA=0:08:09
[05/09 00:20:46 d2.evaluation.evaluator]: Inference done 879/2215. Dataloading: 0.2629 s/iter. Inference: 0.0918 s/iter. Eval: 0.0084 s/iter. Total: 0.3633 s/iter. ETA=0:08:05
[05/09 00:20:51 d2.evaluation.evaluator]: Inference done 893/2215. Dataloading: 0.2631 s/iter. Inference: 0.0914 s/iter. Eval: 0.0084 s/iter. Total: 0.3632 s/iter. ETA=0:08:00
[05/09 00:20:56 d2.evaluation.evaluator]: Inference done 907/2215. Dataloading: 0.2637 s/iter. Inference: 0.0911 s/iter. Eval: 0.0084 s/iter. Total: 0.3634 s/iter. ETA=0:07:55
[05/09 00:21:01 d2.evaluation.evaluator]: Inference done 920/2215. Dataloading: 0.2644 s/iter. Inference: 0.0907 s/iter. Eval: 0.0084 s/iter. Total: 0.3637 s/iter. ETA=0:07:51
[05/09 00:21:06 d2.evaluation.evaluator]: Inference done 933/2215. Dataloading: 0.2652 s/iter. Inference: 0.0904 s/iter. Eval: 0.0084 s/iter. Total: 0.3642 s/iter. ETA=0:07:46
[05/09 00:21:12 d2.evaluation.evaluator]: Inference done 949/2215. Dataloading: 0.2648 s/iter. Inference: 0.0900 s/iter. Eval: 0.0084 s/iter. Total: 0.3635 s/iter. ETA=0:07:40
[05/09 00:21:17 d2.evaluation.evaluator]: Inference done 962/2215. Dataloading: 0.2654 s/iter. Inference: 0.0898 s/iter. Eval: 0.0084 s/iter. Total: 0.3638 s/iter. ETA=0:07:35
[05/09 00:21:22 d2.evaluation.evaluator]: Inference done 977/2215. Dataloading: 0.2655 s/iter. Inference: 0.0895 s/iter. Eval: 0.0085 s/iter. Total: 0.3637 s/iter. ETA=0:07:30
[05/09 00:21:27 d2.evaluation.evaluator]: Inference done 995/2215. Dataloading: 0.2649 s/iter. Inference: 0.0891 s/iter. Eval: 0.0084 s/iter. Total: 0.3626 s/iter. ETA=0:07:22
[05/09 00:21:32 d2.evaluation.evaluator]: Inference done 1010/2215. Dataloading: 0.2648 s/iter. Inference: 0.0887 s/iter. Eval: 0.0084 s/iter. Total: 0.3623 s/iter. ETA=0:07:16
[05/09 00:21:38 d2.evaluation.evaluator]: Inference done 1026/2215. Dataloading: 0.2649 s/iter. Inference: 0.0884 s/iter. Eval: 0.0084 s/iter. Total: 0.3620 s/iter. ETA=0:07:10
[05/09 00:21:43 d2.evaluation.evaluator]: Inference done 1043/2215. Dataloading: 0.2644 s/iter. Inference: 0.0881 s/iter. Eval: 0.0084 s/iter. Total: 0.3610 s/iter. ETA=0:07:03
[05/09 00:21:48 d2.evaluation.evaluator]: Inference done 1059/2215. Dataloading: 0.2641 s/iter. Inference: 0.0878 s/iter. Eval: 0.0084 s/iter. Total: 0.3605 s/iter. ETA=0:06:56
[05/09 00:21:53 d2.evaluation.evaluator]: Inference done 1074/2215. Dataloading: 0.2641 s/iter. Inference: 0.0875 s/iter. Eval: 0.0084 s/iter. Total: 0.3603 s/iter. ETA=0:06:51
[05/09 00:21:59 d2.evaluation.evaluator]: Inference done 1090/2215. Dataloading: 0.2638 s/iter. Inference: 0.0872 s/iter. Eval: 0.0085 s/iter. Total: 0.3596 s/iter. ETA=0:06:44
[05/09 00:22:04 d2.evaluation.evaluator]: Inference done 1106/2215. Dataloading: 0.2637 s/iter. Inference: 0.0869 s/iter. Eval: 0.0084 s/iter. Total: 0.3593 s/iter. ETA=0:06:38
[05/09 00:22:09 d2.evaluation.evaluator]: Inference done 1120/2215. Dataloading: 0.2642 s/iter. Inference: 0.0867 s/iter. Eval: 0.0084 s/iter. Total: 0.3596 s/iter. ETA=0:06:33
[05/09 00:22:15 d2.evaluation.evaluator]: Inference done 1136/2215. Dataloading: 0.2640 s/iter. Inference: 0.0864 s/iter. Eval: 0.0084 s/iter. Total: 0.3591 s/iter. ETA=0:06:27
[05/09 00:22:20 d2.evaluation.evaluator]: Inference done 1150/2215. Dataloading: 0.2649 s/iter. Inference: 0.0862 s/iter. Eval: 0.0084 s/iter. Total: 0.3597 s/iter. ETA=0:06:23
[05/09 00:22:25 d2.evaluation.evaluator]: Inference done 1166/2215. Dataloading: 0.2646 s/iter. Inference: 0.0860 s/iter. Eval: 0.0084 s/iter. Total: 0.3592 s/iter. ETA=0:06:16
[05/09 00:22:31 d2.evaluation.evaluator]: Inference done 1182/2215. Dataloading: 0.2643 s/iter. Inference: 0.0857 s/iter. Eval: 0.0085 s/iter. Total: 0.3587 s/iter. ETA=0:06:10
[05/09 00:22:36 d2.evaluation.evaluator]: Inference done 1196/2215. Dataloading: 0.2645 s/iter. Inference: 0.0855 s/iter. Eval: 0.0085 s/iter. Total: 0.3587 s/iter. ETA=0:06:05
[05/09 00:22:41 d2.evaluation.evaluator]: Inference done 1211/2215. Dataloading: 0.2647 s/iter. Inference: 0.0853 s/iter. Eval: 0.0084 s/iter. Total: 0.3587 s/iter. ETA=0:06:00
[05/09 00:22:46 d2.evaluation.evaluator]: Inference done 1225/2215. Dataloading: 0.2650 s/iter. Inference: 0.0851 s/iter. Eval: 0.0084 s/iter. Total: 0.3587 s/iter. ETA=0:05:55
[05/09 00:22:52 d2.evaluation.evaluator]: Inference done 1240/2215. Dataloading: 0.2654 s/iter. Inference: 0.0849 s/iter. Eval: 0.0084 s/iter. Total: 0.3589 s/iter. ETA=0:05:49
[05/09 00:22:57 d2.evaluation.evaluator]: Inference done 1256/2215. Dataloading: 0.2652 s/iter. Inference: 0.0847 s/iter. Eval: 0.0084 s/iter. Total: 0.3585 s/iter. ETA=0:05:43
[05/09 00:23:02 d2.evaluation.evaluator]: Inference done 1270/2215. Dataloading: 0.2659 s/iter. Inference: 0.0845 s/iter. Eval: 0.0084 s/iter. Total: 0.3590 s/iter. ETA=0:05:39
[05/09 00:23:08 d2.evaluation.evaluator]: Inference done 1286/2215. Dataloading: 0.2656 s/iter. Inference: 0.0843 s/iter. Eval: 0.0084 s/iter. Total: 0.3585 s/iter. ETA=0:05:33
[05/09 00:23:13 d2.evaluation.evaluator]: Inference done 1300/2215. Dataloading: 0.2658 s/iter. Inference: 0.0841 s/iter. Eval: 0.0084 s/iter. Total: 0.3585 s/iter. ETA=0:05:28
[05/09 00:23:18 d2.evaluation.evaluator]: Inference done 1316/2215. Dataloading: 0.2654 s/iter. Inference: 0.0839 s/iter. Eval: 0.0084 s/iter. Total: 0.3580 s/iter. ETA=0:05:21
[05/09 00:23:23 d2.evaluation.evaluator]: Inference done 1332/2215. Dataloading: 0.2651 s/iter. Inference: 0.0838 s/iter. Eval: 0.0084 s/iter. Total: 0.3574 s/iter. ETA=0:05:15
[05/09 00:23:28 d2.evaluation.evaluator]: Inference done 1345/2215. Dataloading: 0.2660 s/iter. Inference: 0.0836 s/iter. Eval: 0.0083 s/iter. Total: 0.3581 s/iter. ETA=0:05:11
[05/09 00:23:33 d2.evaluation.evaluator]: Inference done 1361/2215. Dataloading: 0.2657 s/iter. Inference: 0.0834 s/iter. Eval: 0.0083 s/iter. Total: 0.3577 s/iter. ETA=0:05:05
[05/09 00:23:39 d2.evaluation.evaluator]: Inference done 1375/2215. Dataloading: 0.2660 s/iter. Inference: 0.0833 s/iter. Eval: 0.0084 s/iter. Total: 0.3578 s/iter. ETA=0:05:00
[05/09 00:23:44 d2.evaluation.evaluator]: Inference done 1389/2215. Dataloading: 0.2663 s/iter. Inference: 0.0831 s/iter. Eval: 0.0084 s/iter. Total: 0.3580 s/iter. ETA=0:04:55
[05/09 00:23:49 d2.evaluation.evaluator]: Inference done 1405/2215. Dataloading: 0.2661 s/iter. Inference: 0.0829 s/iter. Eval: 0.0084 s/iter. Total: 0.3576 s/iter. ETA=0:04:49
[05/09 00:23:55 d2.evaluation.evaluator]: Inference done 1419/2215. Dataloading: 0.2666 s/iter. Inference: 0.0828 s/iter. Eval: 0.0084 s/iter. Total: 0.3580 s/iter. ETA=0:04:44
[05/09 00:24:00 d2.evaluation.evaluator]: Inference done 1433/2215. Dataloading: 0.2668 s/iter. Inference: 0.0826 s/iter. Eval: 0.0084 s/iter. Total: 0.3580 s/iter. ETA=0:04:39
[05/09 00:24:05 d2.evaluation.evaluator]: Inference done 1448/2215. Dataloading: 0.2669 s/iter. Inference: 0.0824 s/iter. Eval: 0.0083 s/iter. Total: 0.3579 s/iter. ETA=0:04:34
[05/09 00:24:10 d2.evaluation.evaluator]: Inference done 1463/2215. Dataloading: 0.2670 s/iter. Inference: 0.0823 s/iter. Eval: 0.0083 s/iter. Total: 0.3579 s/iter. ETA=0:04:29
[05/09 00:24:16 d2.evaluation.evaluator]: Inference done 1478/2215. Dataloading: 0.2672 s/iter. Inference: 0.0821 s/iter. Eval: 0.0084 s/iter. Total: 0.3579 s/iter. ETA=0:04:23
[05/09 00:24:21 d2.evaluation.evaluator]: Inference done 1493/2215. Dataloading: 0.2671 s/iter. Inference: 0.0820 s/iter. Eval: 0.0084 s/iter. Total: 0.3577 s/iter. ETA=0:04:18
[05/09 00:24:26 d2.evaluation.evaluator]: Inference done 1506/2215. Dataloading: 0.2677 s/iter. Inference: 0.0819 s/iter. Eval: 0.0084 s/iter. Total: 0.3582 s/iter. ETA=0:04:13
[05/09 00:24:31 d2.evaluation.evaluator]: Inference done 1520/2215. Dataloading: 0.2679 s/iter. Inference: 0.0818 s/iter. Eval: 0.0084 s/iter. Total: 0.3582 s/iter. ETA=0:04:08
[05/09 00:24:36 d2.evaluation.evaluator]: Inference done 1536/2215. Dataloading: 0.2679 s/iter. Inference: 0.0816 s/iter. Eval: 0.0083 s/iter. Total: 0.3580 s/iter. ETA=0:04:03
[05/09 00:24:42 d2.evaluation.evaluator]: Inference done 1552/2215. Dataloading: 0.2677 s/iter. Inference: 0.0815 s/iter. Eval: 0.0083 s/iter. Total: 0.3578 s/iter. ETA=0:03:57
[05/09 00:24:47 d2.evaluation.evaluator]: Inference done 1568/2215. Dataloading: 0.2675 s/iter. Inference: 0.0813 s/iter. Eval: 0.0083 s/iter. Total: 0.3574 s/iter. ETA=0:03:51
[05/09 00:24:53 d2.evaluation.evaluator]: Inference done 1583/2215. Dataloading: 0.2677 s/iter. Inference: 0.0812 s/iter. Eval: 0.0083 s/iter. Total: 0.3575 s/iter. ETA=0:03:45
[05/09 00:24:58 d2.evaluation.evaluator]: Inference done 1597/2215. Dataloading: 0.2679 s/iter. Inference: 0.0811 s/iter. Eval: 0.0083 s/iter. Total: 0.3576 s/iter. ETA=0:03:40
[05/09 00:25:03 d2.evaluation.evaluator]: Inference done 1613/2215. Dataloading: 0.2678 s/iter. Inference: 0.0810 s/iter. Eval: 0.0083 s/iter. Total: 0.3573 s/iter. ETA=0:03:35
[05/09 00:25:08 d2.evaluation.evaluator]: Inference done 1628/2215. Dataloading: 0.2679 s/iter. Inference: 0.0809 s/iter. Eval: 0.0083 s/iter. Total: 0.3573 s/iter. ETA=0:03:29
[05/09 00:25:14 d2.evaluation.evaluator]: Inference done 1644/2215. Dataloading: 0.2677 s/iter. Inference: 0.0808 s/iter. Eval: 0.0083 s/iter. Total: 0.3570 s/iter. ETA=0:03:23
[05/09 00:25:19 d2.evaluation.evaluator]: Inference done 1660/2215. Dataloading: 0.2677 s/iter. Inference: 0.0806 s/iter. Eval: 0.0083 s/iter. Total: 0.3568 s/iter. ETA=0:03:18
[05/09 00:25:25 d2.evaluation.evaluator]: Inference done 1677/2215. Dataloading: 0.2677 s/iter. Inference: 0.0805 s/iter. Eval: 0.0082 s/iter. Total: 0.3567 s/iter. ETA=0:03:11
[05/09 00:25:30 d2.evaluation.evaluator]: Inference done 1692/2215. Dataloading: 0.2677 s/iter. Inference: 0.0804 s/iter. Eval: 0.0083 s/iter. Total: 0.3566 s/iter. ETA=0:03:06
[05/09 00:25:35 d2.evaluation.evaluator]: Inference done 1706/2215. Dataloading: 0.2681 s/iter. Inference: 0.0803 s/iter. Eval: 0.0083 s/iter. Total: 0.3569 s/iter. ETA=0:03:01
[05/09 00:25:40 d2.evaluation.evaluator]: Inference done 1720/2215. Dataloading: 0.2682 s/iter. Inference: 0.0802 s/iter. Eval: 0.0083 s/iter. Total: 0.3569 s/iter. ETA=0:02:56
[05/09 00:25:45 d2.evaluation.evaluator]: Inference done 1733/2215. Dataloading: 0.2685 s/iter. Inference: 0.0801 s/iter. Eval: 0.0083 s/iter. Total: 0.3571 s/iter. ETA=0:02:52
[05/09 00:25:51 d2.evaluation.evaluator]: Inference done 1748/2215. Dataloading: 0.2686 s/iter. Inference: 0.0800 s/iter. Eval: 0.0083 s/iter. Total: 0.3572 s/iter. ETA=0:02:46
[05/09 00:25:56 d2.evaluation.evaluator]: Inference done 1762/2215. Dataloading: 0.2688 s/iter. Inference: 0.0799 s/iter. Eval: 0.0083 s/iter. Total: 0.3572 s/iter. ETA=0:02:41
[05/09 00:26:01 d2.evaluation.evaluator]: Inference done 1774/2215. Dataloading: 0.2693 s/iter. Inference: 0.0798 s/iter. Eval: 0.0083 s/iter. Total: 0.3577 s/iter. ETA=0:02:37
[05/09 00:26:06 d2.evaluation.evaluator]: Inference done 1787/2215. Dataloading: 0.2696 s/iter. Inference: 0.0797 s/iter. Eval: 0.0083 s/iter. Total: 0.3579 s/iter. ETA=0:02:33
[05/09 00:26:12 d2.evaluation.evaluator]: Inference done 1804/2215. Dataloading: 0.2695 s/iter. Inference: 0.0796 s/iter. Eval: 0.0083 s/iter. Total: 0.3576 s/iter. ETA=0:02:26
[05/09 00:26:17 d2.evaluation.evaluator]: Inference done 1820/2215. Dataloading: 0.2694 s/iter. Inference: 0.0795 s/iter. Eval: 0.0083 s/iter. Total: 0.3574 s/iter. ETA=0:02:21
[05/09 00:26:23 d2.evaluation.evaluator]: Inference done 1836/2215. Dataloading: 0.2694 s/iter. Inference: 0.0794 s/iter. Eval: 0.0083 s/iter. Total: 0.3573 s/iter. ETA=0:02:15
[05/09 00:26:28 d2.evaluation.evaluator]: Inference done 1852/2215. Dataloading: 0.2691 s/iter. Inference: 0.0793 s/iter. Eval: 0.0083 s/iter. Total: 0.3569 s/iter. ETA=0:02:09
[05/09 00:26:33 d2.evaluation.evaluator]: Inference done 1867/2215. Dataloading: 0.2691 s/iter. Inference: 0.0792 s/iter. Eval: 0.0082 s/iter. Total: 0.3568 s/iter. ETA=0:02:04
[05/09 00:26:38 d2.evaluation.evaluator]: Inference done 1881/2215. Dataloading: 0.2693 s/iter. Inference: 0.0791 s/iter. Eval: 0.0082 s/iter. Total: 0.3569 s/iter. ETA=0:01:59
[05/09 00:26:43 d2.evaluation.evaluator]: Inference done 1893/2215. Dataloading: 0.2700 s/iter. Inference: 0.0791 s/iter. Eval: 0.0082 s/iter. Total: 0.3575 s/iter. ETA=0:01:55
[05/09 00:26:48 d2.evaluation.evaluator]: Inference done 1907/2215. Dataloading: 0.2701 s/iter. Inference: 0.0790 s/iter. Eval: 0.0083 s/iter. Total: 0.3575 s/iter. ETA=0:01:50
[05/09 00:26:54 d2.evaluation.evaluator]: Inference done 1921/2215. Dataloading: 0.2704 s/iter. Inference: 0.0789 s/iter. Eval: 0.0083 s/iter. Total: 0.3578 s/iter. ETA=0:01:45
[05/09 00:26:59 d2.evaluation.evaluator]: Inference done 1936/2215. Dataloading: 0.2705 s/iter. Inference: 0.0788 s/iter. Eval: 0.0082 s/iter. Total: 0.3578 s/iter. ETA=0:01:39
[05/09 00:27:05 d2.evaluation.evaluator]: Inference done 1951/2215. Dataloading: 0.2709 s/iter. Inference: 0.0787 s/iter. Eval: 0.0083 s/iter. Total: 0.3581 s/iter. ETA=0:01:34
[05/09 00:27:10 d2.evaluation.evaluator]: Inference done 1965/2215. Dataloading: 0.2710 s/iter. Inference: 0.0787 s/iter. Eval: 0.0083 s/iter. Total: 0.3582 s/iter. ETA=0:01:29
[05/09 00:27:16 d2.evaluation.evaluator]: Inference done 1981/2215. Dataloading: 0.2708 s/iter. Inference: 0.0786 s/iter. Eval: 0.0083 s/iter. Total: 0.3579 s/iter. ETA=0:01:23
[05/09 00:27:21 d2.evaluation.evaluator]: Inference done 1995/2215. Dataloading: 0.2709 s/iter. Inference: 0.0785 s/iter. Eval: 0.0083 s/iter. Total: 0.3579 s/iter. ETA=0:01:18
[05/09 00:27:27 d2.evaluation.evaluator]: Inference done 2011/2215. Dataloading: 0.2712 s/iter. Inference: 0.0784 s/iter. Eval: 0.0082 s/iter. Total: 0.3581 s/iter. ETA=0:01:13
[05/09 00:27:32 d2.evaluation.evaluator]: Inference done 2025/2215. Dataloading: 0.2713 s/iter. Inference: 0.0784 s/iter. Eval: 0.0082 s/iter. Total: 0.3581 s/iter. ETA=0:01:08
[05/09 00:27:37 d2.evaluation.evaluator]: Inference done 2039/2215. Dataloading: 0.2715 s/iter. Inference: 0.0783 s/iter. Eval: 0.0082 s/iter. Total: 0.3583 s/iter. ETA=0:01:03
[05/09 00:27:42 d2.evaluation.evaluator]: Inference done 2053/2215. Dataloading: 0.2717 s/iter. Inference: 0.0782 s/iter. Eval: 0.0082 s/iter. Total: 0.3583 s/iter. ETA=0:00:58
[05/09 00:27:48 d2.evaluation.evaluator]: Inference done 2069/2215. Dataloading: 0.2717 s/iter. Inference: 0.0781 s/iter. Eval: 0.0082 s/iter. Total: 0.3582 s/iter. ETA=0:00:52
[05/09 00:27:53 d2.evaluation.evaluator]: Inference done 2085/2215. Dataloading: 0.2716 s/iter. Inference: 0.0780 s/iter. Eval: 0.0082 s/iter. Total: 0.3581 s/iter. ETA=0:00:46
[05/09 00:27:58 d2.evaluation.evaluator]: Inference done 2100/2215. Dataloading: 0.2715 s/iter. Inference: 0.0780 s/iter. Eval: 0.0082 s/iter. Total: 0.3579 s/iter. ETA=0:00:41
[05/09 00:28:04 d2.evaluation.evaluator]: Inference done 2114/2215. Dataloading: 0.2722 s/iter. Inference: 0.0779 s/iter. Eval: 0.0082 s/iter. Total: 0.3585 s/iter. ETA=0:00:36
[05/09 00:28:10 d2.evaluation.evaluator]: Inference done 2130/2215. Dataloading: 0.2721 s/iter. Inference: 0.0778 s/iter. Eval: 0.0082 s/iter. Total: 0.3584 s/iter. ETA=0:00:30
[05/09 00:28:15 d2.evaluation.evaluator]: Inference done 2144/2215. Dataloading: 0.2724 s/iter. Inference: 0.0778 s/iter. Eval: 0.0081 s/iter. Total: 0.3585 s/iter. ETA=0:00:25
[05/09 00:28:20 d2.evaluation.evaluator]: Inference done 2158/2215. Dataloading: 0.2725 s/iter. Inference: 0.0777 s/iter. Eval: 0.0081 s/iter. Total: 0.3586 s/iter. ETA=0:00:20
[05/09 00:28:25 d2.evaluation.evaluator]: Inference done 2171/2215. Dataloading: 0.2728 s/iter. Inference: 0.0776 s/iter. Eval: 0.0081 s/iter. Total: 0.3588 s/iter. ETA=0:00:15
[05/09 00:28:31 d2.evaluation.evaluator]: Inference done 2186/2215. Dataloading: 0.2728 s/iter. Inference: 0.0776 s/iter. Eval: 0.0081 s/iter. Total: 0.3587 s/iter. ETA=0:00:10
[05/09 00:28:36 d2.evaluation.evaluator]: Inference done 2202/2215. Dataloading: 0.2727 s/iter. Inference: 0.0775 s/iter. Eval: 0.0082 s/iter. Total: 0.3586 s/iter. ETA=0:00:04
[05/09 00:28:41 d2.evaluation.evaluator]: Total inference time: 0:13:12.528540 (0.358610 s / iter per device, on 1 devices)
[05/09 00:28:41 d2.evaluation.evaluator]: Total inference pure compute time: 0:02:51 (0.077470 s / iter per device, on 1 devices)
[05/09 00:28:41 d2.evaluation.coco_evaluation]: Preparing results for COCO format ...
[05/09 00:28:41 d2.evaluation.coco_evaluation]: Saving results to ./output_evals/coco_instances_results.json
[05/09 00:28:42 d2.evaluation.coco_evaluation]: Evaluating predictions with unofficial COCO API...
Loading and preparing results...
DONE (t=0.03s)
creating index...
index created!
[05/09 00:28:42 d2.evaluation.fast_eval_api]: Evaluate annotation type *bbox*
[05/09 00:28:42 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 0.73 seconds.
[05/09 00:28:42 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 00:28:42 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.12 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.121
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.192
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.134
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.170
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.120
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.121
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.214
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.215
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.293
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.215
[05/09 00:28:42 d2.evaluation.coco_evaluation]: Evaluation results for bbox: 
|   AP   |  AP50  |  AP75  |  APs   |  APm   |  APl   |
|:------:|:------:|:------:|:------:|:------:|:------:|
| 12.055 | 19.171 | 13.449 | 50.000 | 16.957 | 11.992 |
[05/09 00:28:42 d2.evaluation.coco_evaluation]: Per-category bbox AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 40.045 | carry      | 0.000 | point      | 0.000  |
| eat           | 1.653  | drink      | 0.000 | stand      | 42.685 |
| talk_on_phone | 0.000  |            |       |            |        |
Loading and preparing results...
DONE (t=0.34s)
creating index...
index created!
[05/09 00:28:44 d2.evaluation.fast_eval_api]: Evaluate annotation type *segm*
[05/09 00:28:45 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 1.38 seconds.
[05/09 00:28:45 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 00:28:45 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.12 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.112
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.186
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.126
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.033
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.120
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.118
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.209
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.210
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.400
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.282
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.212
[05/09 00:28:45 d2.evaluation.coco_evaluation]: Evaluation results for segm: 
|   AP   |  AP50  |  AP75  |  APs  |  APm   |  APl   |
|:------:|:------:|:------:|:-----:|:------:|:------:|
| 11.216 | 18.607 | 12.625 | 3.333 | 12.010 | 12.547 |
[05/09 00:28:45 d2.evaluation.coco_evaluation]: Per-category segm AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 34.787 | carry      | 0.000 | point      | 0.000  |
| eat           | 2.034  | drink      | 0.000 | stand      | 41.690 |
| talk_on_phone | 0.000  |            |       |            |        |

OrderedDict([('bbox',
              {'AP': 12.054746090751948,
               'AP-carry': 0.0,
               'AP-drink': 0.0,
               'AP-eat': 1.653387355812205,
               'AP-hold': 40.04496914199544,
               'AP-point': 0.0,
               'AP-stand': 42.684866137455984,
               'AP-talk_on_phone': 0.0,
               'AP50': 19.171166864176435,
               'AP75': 13.44908209568981,
               'APl': 11.992233646442443,
               'APm': 16.956725039905564,
               'APs': 50.0}),
             ('segm',
              {'AP': 11.215705604418499,
               'AP-carry': 0.0,
               'AP-drink': 0.0,
               'AP-eat': 2.033828393624494,
               'AP-hold': 34.786599491454645,
               'AP-point': 0.0,
               'AP-stand': 41.68951134585036,
               'AP-talk_on_phone': 0.0,
               'AP50': 18.607357993512416,
               'AP75': 12.624930670282072,
               'APl': 12.54704110216769,
               'APm': 12.010120116297184,
               'APs': 3.3333333333333326})])

Run Eval Metrics for 0.4

cfg.MODEL.WEIGHTS = os.path.join(cfg.OUTPUT_DIR, "model_final.pth")
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.4   # set the testing threshold for this model
cfg.DATASETS.TEST = ("rel_actions_val3", )
predictor = DefaultPredictor(cfg)

[05/09 00:31:05 d2.checkpoint.c2_model_loading]: Following weights matched with model:
| Names in Model                                  | Names in Checkpoint                                                                                  | Shapes                                          |
|:------------------------------------------------|:-----------------------------------------------------------------------------------------------------|:------------------------------------------------|
| backbone.bottom_up.res2.0.conv1.*               | backbone.bottom_up.res2.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,1,1)             |
| backbone.bottom_up.res2.0.conv2.*               | backbone.bottom_up.res2.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.0.conv3.*               | backbone.bottom_up.res2.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.0.shortcut.*            | backbone.bottom_up.res2.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.1.conv1.*               | backbone.bottom_up.res2.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.1.conv2.*               | backbone.bottom_up.res2.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.1.conv3.*               | backbone.bottom_up.res2.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.2.conv1.*               | backbone.bottom_up.res2.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.2.conv2.*               | backbone.bottom_up.res2.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.2.conv3.*               | backbone.bottom_up.res2.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res3.0.conv1.*               | backbone.bottom_up.res3.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,256,1,1)       |
| backbone.bottom_up.res3.0.conv2.*               | backbone.bottom_up.res3.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.0.conv3.*               | backbone.bottom_up.res3.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.0.shortcut.*            | backbone.bottom_up.res3.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (512,) (512,) (512,) (512,) (512,256,1,1)       |
| backbone.bottom_up.res3.1.conv1.*               | backbone.bottom_up.res3.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.1.conv2.*               | backbone.bottom_up.res3.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.1.conv3.*               | backbone.bottom_up.res3.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.2.conv1.*               | backbone.bottom_up.res3.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.2.conv2.*               | backbone.bottom_up.res3.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.2.conv3.*               | backbone.bottom_up.res3.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.3.conv1.*               | backbone.bottom_up.res3.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.3.conv2.*               | backbone.bottom_up.res3.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.3.conv3.*               | backbone.bottom_up.res3.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res4.0.conv1.*               | backbone.bottom_up.res4.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,512,1,1)       |
| backbone.bottom_up.res4.0.conv2.*               | backbone.bottom_up.res4.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.0.conv3.*               | backbone.bottom_up.res4.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.0.shortcut.*            | backbone.bottom_up.res4.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (1024,) (1024,) (1024,) (1024,) (1024,512,1,1)  |
| backbone.bottom_up.res4.1.conv1.*               | backbone.bottom_up.res4.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.1.conv2.*               | backbone.bottom_up.res4.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.1.conv3.*               | backbone.bottom_up.res4.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.2.conv1.*               | backbone.bottom_up.res4.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.2.conv2.*               | backbone.bottom_up.res4.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.2.conv3.*               | backbone.bottom_up.res4.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.3.conv1.*               | backbone.bottom_up.res4.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.3.conv2.*               | backbone.bottom_up.res4.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.3.conv3.*               | backbone.bottom_up.res4.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.4.conv1.*               | backbone.bottom_up.res4.4.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.4.conv2.*               | backbone.bottom_up.res4.4.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.4.conv3.*               | backbone.bottom_up.res4.4.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.5.conv1.*               | backbone.bottom_up.res4.5.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.5.conv2.*               | backbone.bottom_up.res4.5.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.5.conv3.*               | backbone.bottom_up.res4.5.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res5.0.conv1.*               | backbone.bottom_up.res5.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,1024,1,1)      |
| backbone.bottom_up.res5.0.conv2.*               | backbone.bottom_up.res5.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.0.conv3.*               | backbone.bottom_up.res5.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.0.shortcut.*            | backbone.bottom_up.res5.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (2048,) (2048,) (2048,) (2048,) (2048,1024,1,1) |
| backbone.bottom_up.res5.1.conv1.*               | backbone.bottom_up.res5.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.1.conv2.*               | backbone.bottom_up.res5.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.1.conv3.*               | backbone.bottom_up.res5.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.2.conv1.*               | backbone.bottom_up.res5.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.2.conv2.*               | backbone.bottom_up.res5.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.2.conv3.*               | backbone.bottom_up.res5.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.stem.conv1.*                 | backbone.bottom_up.stem.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}      | (64,) (64,) (64,) (64,) (64,3,7,7)              |
| backbone.fpn_lateral2.*                         | backbone.fpn_lateral2.{bias,weight}                                                                  | (256,) (256,256,1,1)                            |
| backbone.fpn_lateral3.*                         | backbone.fpn_lateral3.{bias,weight}                                                                  | (256,) (256,512,1,1)                            |
| backbone.fpn_lateral4.*                         | backbone.fpn_lateral4.{bias,weight}                                                                  | (256,) (256,1024,1,1)                           |
| backbone.fpn_lateral5.*                         | backbone.fpn_lateral5.{bias,weight}                                                                  | (256,) (256,2048,1,1)                           |
| backbone.fpn_output2.*                          | backbone.fpn_output2.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output3.*                          | backbone.fpn_output3.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output4.*                          | backbone.fpn_output4.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output5.*                          | backbone.fpn_output5.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.anchor_deltas.*     | proposal_generator.rpn_head.anchor_deltas.{bias,weight}                                              | (12,) (12,256,1,1)                              |
| proposal_generator.rpn_head.conv.*              | proposal_generator.rpn_head.conv.{bias,weight}                                                       | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.objectness_logits.* | proposal_generator.rpn_head.objectness_logits.{bias,weight}                                          | (3,) (3,256,1,1)                                |
| roi_heads.box_head.fc1.*                        | roi_heads.box_head.fc1.{bias,weight}                                                                 | (1024,) (1024,12544)                            |
| roi_heads.box_head.fc2.*                        | roi_heads.box_head.fc2.{bias,weight}                                                                 | (1024,) (1024,1024)                             |
| roi_heads.box_predictor.bbox_pred.*             | roi_heads.box_predictor.bbox_pred.{bias,weight}                                                      | (28,) (28,1024)                                 |
| roi_heads.box_predictor.cls_score.*             | roi_heads.box_predictor.cls_score.{bias,weight}                                                      | (8,) (8,1024)                                   |
| roi_heads.mask_head.deconv.*                    | roi_heads.mask_head.deconv.{bias,weight}                                                             | (256,) (256,256,2,2)                            |
| roi_heads.mask_head.mask_fcn1.*                 | roi_heads.mask_head.mask_fcn1.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn2.*                 | roi_heads.mask_head.mask_fcn2.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn3.*                 | roi_heads.mask_head.mask_fcn3.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.mask_fcn4.*                 | roi_heads.mask_head.mask_fcn4.{bias,weight}                                                          | (256,) (256,256,3,3)                            |
| roi_heads.mask_head.predictor.*                 | roi_heads.mask_head.predictor.{bias,weight}                                                          | (7,) (7,256,1,1)                                |

from detectron2.evaluation import COCOEvaluator, inference_on_dataset
from detectron2.data import build_detection_test_loader
from detectron2.engine import DefaultTrainer

evaluator = COCOEvaluator("rel_actions_val3", cfg, False, output_dir="./output_evals/")
val_loader = build_detection_test_loader(cfg, "rel_actions_val3")
inference_on_dataset(trainer.model, val_loader, evaluator)

WARNING [05/09 00:32:15 d2.evaluation.coco_evaluation]: COCO Evaluator instantiated using config, this is deprecated behavior. Please pass in explicit arguments instead.
[05/09 00:32:15 d2.data.datasets.coco]: Loaded 2215 images in COCO format from ./data/instances_vcoco_actions_val_2014.json
[05/09 00:32:15 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in inference: [ResizeShortestEdge(short_edge_length=(800, 800), max_size=1333, sample_style='choice')]
[05/09 00:32:15 d2.data.common]: Serializing 2215 elements to byte tensors and concatenating them all ...
[05/09 00:32:15 d2.data.common]: Serialized dataset takes 5.66 MiB
[05/09 00:32:15 d2.evaluation.evaluator]: Start inference on 2215 batches
[05/09 00:32:16 d2.evaluation.evaluator]: Inference done 11/2215. Dataloading: 0.0017 s/iter. Inference: 0.0679 s/iter. Eval: 0.0063 s/iter. Total: 0.0759 s/iter. ETA=0:02:47
[05/09 00:32:21 d2.evaluation.evaluator]: Inference done 73/2215. Dataloading: 0.0022 s/iter. Inference: 0.0702 s/iter. Eval: 0.0078 s/iter. Total: 0.0804 s/iter. ETA=0:02:52
[05/09 00:32:27 d2.evaluation.evaluator]: Inference done 135/2215. Dataloading: 0.0022 s/iter. Inference: 0.0698 s/iter. Eval: 0.0086 s/iter. Total: 0.0808 s/iter. ETA=0:02:48
[05/09 00:32:32 d2.evaluation.evaluator]: Inference done 197/2215. Dataloading: 0.0022 s/iter. Inference: 0.0700 s/iter. Eval: 0.0086 s/iter. Total: 0.0809 s/iter. ETA=0:02:43
[05/09 00:32:37 d2.evaluation.evaluator]: Inference done 260/2215. Dataloading: 0.0022 s/iter. Inference: 0.0700 s/iter. Eval: 0.0084 s/iter. Total: 0.0807 s/iter. ETA=0:02:37
[05/09 00:32:42 d2.evaluation.evaluator]: Inference done 322/2215. Dataloading: 0.0022 s/iter. Inference: 0.0700 s/iter. Eval: 0.0084 s/iter. Total: 0.0808 s/iter. ETA=0:02:32
[05/09 00:32:47 d2.evaluation.evaluator]: Inference done 385/2215. Dataloading: 0.0022 s/iter. Inference: 0.0701 s/iter. Eval: 0.0083 s/iter. Total: 0.0807 s/iter. ETA=0:02:27
[05/09 00:32:52 d2.evaluation.evaluator]: Inference done 448/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0081 s/iter. Total: 0.0806 s/iter. ETA=0:02:22
[05/09 00:32:57 d2.evaluation.evaluator]: Inference done 511/2215. Dataloading: 0.0023 s/iter. Inference: 0.0700 s/iter. Eval: 0.0082 s/iter. Total: 0.0806 s/iter. ETA=0:02:17
[05/09 00:33:02 d2.evaluation.evaluator]: Inference done 574/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0081 s/iter. Total: 0.0806 s/iter. ETA=0:02:12
[05/09 00:33:07 d2.evaluation.evaluator]: Inference done 636/2215. Dataloading: 0.0023 s/iter. Inference: 0.0700 s/iter. Eval: 0.0082 s/iter. Total: 0.0806 s/iter. ETA=0:02:07
[05/09 00:33:12 d2.evaluation.evaluator]: Inference done 697/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:02:02
[05/09 00:33:17 d2.evaluation.evaluator]: Inference done 758/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0083 s/iter. Total: 0.0809 s/iter. ETA=0:01:57
[05/09 00:33:22 d2.evaluation.evaluator]: Inference done 820/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0083 s/iter. Total: 0.0810 s/iter. ETA=0:01:52
[05/09 00:33:27 d2.evaluation.evaluator]: Inference done 884/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:47
[05/09 00:33:32 d2.evaluation.evaluator]: Inference done 947/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:42
[05/09 00:33:37 d2.evaluation.evaluator]: Inference done 1009/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:37
[05/09 00:33:42 d2.evaluation.evaluator]: Inference done 1072/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:32
[05/09 00:33:47 d2.evaluation.evaluator]: Inference done 1134/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:27
[05/09 00:33:52 d2.evaluation.evaluator]: Inference done 1196/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0082 s/iter. Total: 0.0808 s/iter. ETA=0:01:22
[05/09 00:33:57 d2.evaluation.evaluator]: Inference done 1259/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0081 s/iter. Total: 0.0807 s/iter. ETA=0:01:17
[05/09 00:34:02 d2.evaluation.evaluator]: Inference done 1322/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0081 s/iter. Total: 0.0807 s/iter. ETA=0:01:12
[05/09 00:34:07 d2.evaluation.evaluator]: Inference done 1384/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0081 s/iter. Total: 0.0807 s/iter. ETA=0:01:07
[05/09 00:34:12 d2.evaluation.evaluator]: Inference done 1448/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0081 s/iter. Total: 0.0806 s/iter. ETA=0:01:01
[05/09 00:34:17 d2.evaluation.evaluator]: Inference done 1510/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0081 s/iter. Total: 0.0806 s/iter. ETA=0:00:56
[05/09 00:34:22 d2.evaluation.evaluator]: Inference done 1575/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0080 s/iter. Total: 0.0805 s/iter. ETA=0:00:51
[05/09 00:34:27 d2.evaluation.evaluator]: Inference done 1637/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0805 s/iter. ETA=0:00:46
[05/09 00:34:33 d2.evaluation.evaluator]: Inference done 1699/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0806 s/iter. ETA=0:00:41
[05/09 00:34:38 d2.evaluation.evaluator]: Inference done 1762/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0806 s/iter. ETA=0:00:36
[05/09 00:34:43 d2.evaluation.evaluator]: Inference done 1825/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0806 s/iter. ETA=0:00:31
[05/09 00:34:48 d2.evaluation.evaluator]: Inference done 1889/2215. Dataloading: 0.0023 s/iter. Inference: 0.0701 s/iter. Eval: 0.0079 s/iter. Total: 0.0805 s/iter. ETA=0:00:26
[05/09 00:34:53 d2.evaluation.evaluator]: Inference done 1949/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0806 s/iter. ETA=0:00:21
[05/09 00:34:58 d2.evaluation.evaluator]: Inference done 2011/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0080 s/iter. Total: 0.0806 s/iter. ETA=0:00:16
[05/09 00:35:03 d2.evaluation.evaluator]: Inference done 2075/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0079 s/iter. Total: 0.0805 s/iter. ETA=0:00:11
[05/09 00:35:08 d2.evaluation.evaluator]: Inference done 2138/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0079 s/iter. Total: 0.0805 s/iter. ETA=0:00:06
[05/09 00:35:13 d2.evaluation.evaluator]: Inference done 2200/2215. Dataloading: 0.0023 s/iter. Inference: 0.0702 s/iter. Eval: 0.0079 s/iter. Total: 0.0805 s/iter. ETA=0:00:01
[05/09 00:35:14 d2.evaluation.evaluator]: Total inference time: 0:02:58.067442 (0.080574 s / iter per device, on 1 devices)
[05/09 00:35:14 d2.evaluation.evaluator]: Total inference pure compute time: 0:02:35 (0.070239 s / iter per device, on 1 devices)
[05/09 00:35:14 d2.evaluation.coco_evaluation]: Preparing results for COCO format ...
[05/09 00:35:14 d2.evaluation.coco_evaluation]: Saving results to ./output_evals/coco_instances_results.json
[05/09 00:35:15 d2.evaluation.coco_evaluation]: Evaluating predictions with unofficial COCO API...
Loading and preparing results...
DONE (t=0.04s)
creating index...
index created!
[05/09 00:35:15 d2.evaluation.fast_eval_api]: Evaluate annotation type *bbox*
[05/09 00:35:16 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 1.21 seconds.
[05/09 00:35:16 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 00:35:16 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.12 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.121
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.192
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.134
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.170
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.120
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.121
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.214
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.215
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.293
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.215
[05/09 00:35:16 d2.evaluation.coco_evaluation]: Evaluation results for bbox: 
|   AP   |  AP50  |  AP75  |  APs   |  APm   |  APl   |
|:------:|:------:|:------:|:------:|:------:|:------:|
| 12.055 | 19.171 | 13.449 | 50.000 | 16.957 | 11.992 |
[05/09 00:35:16 d2.evaluation.coco_evaluation]: Per-category bbox AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 40.045 | carry      | 0.000 | point      | 0.000  |
| eat           | 1.653  | drink      | 0.000 | stand      | 42.685 |
| talk_on_phone | 0.000  |            |       |            |        |
Loading and preparing results...
DONE (t=0.36s)
creating index...
index created!
[05/09 00:35:17 d2.evaluation.fast_eval_api]: Evaluate annotation type *segm*
[05/09 00:35:18 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 1.44 seconds.
[05/09 00:35:18 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 00:35:19 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.12 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.112
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.186
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.126
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.033
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.120
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.125
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.118
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.209
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.210
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.400
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.282
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.212
[05/09 00:35:19 d2.evaluation.coco_evaluation]: Evaluation results for segm: 
|   AP   |  AP50  |  AP75  |  APs  |  APm   |  APl   |
|:------:|:------:|:------:|:-----:|:------:|:------:|
| 11.216 | 18.607 | 12.625 | 3.333 | 12.010 | 12.547 |
[05/09 00:35:19 d2.evaluation.coco_evaluation]: Per-category segm AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 34.787 | carry      | 0.000 | point      | 0.000  |
| eat           | 2.034  | drink      | 0.000 | stand      | 41.690 |
| talk_on_phone | 0.000  |            |       |            |        |

OrderedDict([('bbox',
              {'AP': 12.054746090751948,
               'AP-carry': 0.0,
               'AP-drink': 0.0,
               'AP-eat': 1.653387355812205,
               'AP-hold': 40.04496914199544,
               'AP-point': 0.0,
               'AP-stand': 42.684866137455984,
               'AP-talk_on_phone': 0.0,
               'AP50': 19.171166864176435,
               'AP75': 13.44908209568981,
               'APl': 11.992233646442443,
               'APm': 16.956725039905564,
               'APs': 50.0}),
             ('segm',
              {'AP': 11.215705604418499,
               'AP-carry': 0.0,
               'AP-drink': 0.0,
               'AP-eat': 2.033828393624494,
               'AP-hold': 34.786599491454645,
               'AP-point': 0.0,
               'AP-stand': 41.68951134585036,
               'AP-talk_on_phone': 0.0,
               'AP50': 18.607357993512416,
               'AP75': 12.624930670282072,
               'APl': 12.54704110216769,
               'APm': 12.010120116297184,
               'APs': 3.3333333333333326})])

Train on Faster RCNN Model

COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml

from detectron2.engine import DefaultTrainer
from detectron2.config import get_cfg
import os

from detectron2 import model_zoo

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file("COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml"))
cfg.DATASETS.TRAIN = ("rel_actions1",)
cfg.DATASETS.TEST = ()   # no metrics implemented for this dataset
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url("COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml")  # initialize from model zoo
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.SOLVER.BASE_LR = 0.002
cfg.SOLVER.MAX_ITER = 1000    
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 128   # faster, and good enough for this toy dataset
cfg.MODEL.ROI_HEADS.NUM_CLASSES = len(relevant_classes)  # all relevant classes
cfg.OUTPUT_DIR = "./outputs/action_only_detector"

os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
trainer = DefaultTrainer(cfg)
trainer.resume_or_load(resume=False)
trainer.train()

[05/09 00:48:15 d2.engine.defaults]: Model:
GeneralizedRCNN(
  (backbone): FPN(
    (fpn_lateral2): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral3): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral4): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output4): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral5): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output5): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (top_block): LastLevelMaxPool()
    (bottom_up): ResNet(
      (stem): BasicStem(
        (conv1): Conv2d(
          3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False
          (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
        )
      )
      (res2): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv1): Conv2d(
            64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
      )
      (res3): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv1): Conv2d(
            256, 128, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
      )
      (res4): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
          (conv1): Conv2d(
            512, 256, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (4): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (5): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
      )
      (res5): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
          (conv1): Conv2d(
            1024, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
      )
    )
  )
  (proposal_generator): RPN(
    (rpn_head): StandardRPNHead(
      (conv): Conv2d(
        256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)
        (activation): ReLU()
      )
      (objectness_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (anchor_deltas): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
    (anchor_generator): DefaultAnchorGenerator(
      (cell_anchors): BufferList()
    )
  )
  (roi_heads): StandardROIHeads(
    (box_pooler): ROIPooler(
      (level_poolers): ModuleList(
        (0): ROIAlign(output_size=(7, 7), spatial_scale=0.25, sampling_ratio=0, aligned=True)
        (1): ROIAlign(output_size=(7, 7), spatial_scale=0.125, sampling_ratio=0, aligned=True)
        (2): ROIAlign(output_size=(7, 7), spatial_scale=0.0625, sampling_ratio=0, aligned=True)
        (3): ROIAlign(output_size=(7, 7), spatial_scale=0.03125, sampling_ratio=0, aligned=True)
      )
    )
    (box_head): FastRCNNConvFCHead(
      (flatten): Flatten(start_dim=1, end_dim=-1)
      (fc1): Linear(in_features=12544, out_features=1024, bias=True)
      (fc_relu1): ReLU()
      (fc2): Linear(in_features=1024, out_features=1024, bias=True)
      (fc_relu2): ReLU()
    )
    (box_predictor): FastRCNNOutputLayers(
      (cls_score): Linear(in_features=1024, out_features=8, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=28, bias=True)
    )
  )
)
[05/09 00:48:15 d2.data.datasets.coco]: Loaded 1951 images in COCO format from ./data/instances_vcoco_actions_train.json
[05/09 00:48:15 d2.data.build]: Removed 0 images with no usable annotations. 1951 images left.
[05/09 00:48:15 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in training: [ResizeShortestEdge(short_edge_length=(640, 672, 704, 736, 768, 800), max_size=1333, sample_style='choice'), RandomFlip()]
[05/09 00:48:15 d2.data.build]: Using training sampler TrainingSampler
[05/09 00:48:15 d2.data.common]: Serializing 1951 elements to byte tensors and concatenating them all ...
[05/09 00:48:15 d2.data.common]: Serialized dataset takes 4.90 MiB
WARNING [05/09 00:48:15 d2.solver.build]: SOLVER.STEPS contains values larger than SOLVER.MAX_ITER. These values will be ignored.

model_final_280758.pkl: 167MB [00:20, 8.06MB/s]                           
Skip loading parameter 'roi_heads.box_predictor.cls_score.weight' to the model due to incompatible shapes: (81, 1024) in the checkpoint but (8, 1024) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.cls_score.bias' to the model due to incompatible shapes: (81,) in the checkpoint but (8,) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.bbox_pred.weight' to the model due to incompatible shapes: (320, 1024) in the checkpoint but (28, 1024) in the model! You might want to double check if this is expected.
Skip loading parameter 'roi_heads.box_predictor.bbox_pred.bias' to the model due to incompatible shapes: (320,) in the checkpoint but (28,) in the model! You might want to double check if this is expected.
Some model parameters or buffers are not found in the checkpoint:
roi_heads.box_predictor.bbox_pred.{bias, weight}
roi_heads.box_predictor.cls_score.{bias, weight}

[05/09 00:48:36 d2.engine.train_loop]: Starting training from iteration 0
[05/09 00:48:42 d2.utils.events]:  eta: 0:04:03  iter: 19  total_loss: 2.62  loss_cls: 2.017  loss_box_reg: 0.5686  loss_rpn_cls: 0.0201  loss_rpn_loc: 0.00475  time: 0.2523  data_time: 0.0253  lr: 3.9962e-05  max_mem: 2758M
[05/09 00:48:47 d2.utils.events]:  eta: 0:04:04  iter: 39  total_loss: 2.04  loss_cls: 1.369  loss_box_reg: 0.6559  loss_rpn_cls: 0.02173  loss_rpn_loc: 0.00415  time: 0.2630  data_time: 0.0128  lr: 7.9922e-05  max_mem: 2806M
[05/09 00:48:53 d2.utils.events]:  eta: 0:04:03  iter: 59  total_loss: 1.361  loss_cls: 0.7126  loss_box_reg: 0.5817  loss_rpn_cls: 0.01871  loss_rpn_loc: 0.005755  time: 0.2667  data_time: 0.0162  lr: 0.00011988  max_mem: 2806M
[05/09 00:48:58 d2.utils.events]:  eta: 0:04:00  iter: 79  total_loss: 1.053  loss_cls: 0.4737  loss_box_reg: 0.5136  loss_rpn_cls: 0.01666  loss_rpn_loc: 0.003539  time: 0.2644  data_time: 0.0147  lr: 0.00015984  max_mem: 2806M
[05/09 00:49:04 d2.utils.events]:  eta: 0:03:59  iter: 99  total_loss: 1.024  loss_cls: 0.4274  loss_box_reg: 0.5611  loss_rpn_cls: 0.01437  loss_rpn_loc: 0.004409  time: 0.2771  data_time: 0.0127  lr: 0.0001998  max_mem: 2895M
[05/09 00:49:10 d2.utils.events]:  eta: 0:03:54  iter: 119  total_loss: 1.034  loss_cls: 0.4182  loss_box_reg: 0.5902  loss_rpn_cls: 0.01133  loss_rpn_loc: 0.004913  time: 0.2783  data_time: 0.0126  lr: 0.00023976  max_mem: 2895M
[05/09 00:49:15 d2.utils.events]:  eta: 0:03:46  iter: 139  total_loss: 0.9363  loss_cls: 0.3477  loss_box_reg: 0.5706  loss_rpn_cls: 0.01248  loss_rpn_loc: 0.003268  time: 0.2744  data_time: 0.0135  lr: 0.00027972  max_mem: 2895M
[05/09 00:49:21 d2.utils.events]:  eta: 0:03:43  iter: 159  total_loss: 0.7916  loss_cls: 0.285  loss_box_reg: 0.4965  loss_rpn_cls: 0.005246  loss_rpn_loc: 0.005301  time: 0.2744  data_time: 0.0148  lr: 0.00031968  max_mem: 2895M
[05/09 00:49:26 d2.utils.events]:  eta: 0:03:38  iter: 179  total_loss: 0.8213  loss_cls: 0.2753  loss_box_reg: 0.5065  loss_rpn_cls: 0.005218  loss_rpn_loc: 0.005492  time: 0.2733  data_time: 0.0123  lr: 0.00035964  max_mem: 2895M
[05/09 00:49:31 d2.utils.events]:  eta: 0:03:32  iter: 199  total_loss: 0.5861  loss_cls: 0.1765  loss_box_reg: 0.3641  loss_rpn_cls: 0.00506  loss_rpn_loc: 0.005295  time: 0.2726  data_time: 0.0107  lr: 0.0003996  max_mem: 2895M
[05/09 00:49:37 d2.utils.events]:  eta: 0:03:27  iter: 219  total_loss: 0.6405  loss_cls: 0.2402  loss_box_reg: 0.3636  loss_rpn_cls: 0.003551  loss_rpn_loc: 0.005702  time: 0.2724  data_time: 0.0153  lr: 0.00043956  max_mem: 2895M
[05/09 00:49:42 d2.utils.events]:  eta: 0:03:21  iter: 239  total_loss: 0.5892  loss_cls: 0.2492  loss_box_reg: 0.3051  loss_rpn_cls: 0.004994  loss_rpn_loc: 0.007571  time: 0.2719  data_time: 0.0121  lr: 0.00047952  max_mem: 2895M
[05/09 00:49:47 d2.utils.events]:  eta: 0:03:14  iter: 259  total_loss: 0.506  loss_cls: 0.2352  loss_box_reg: 0.2684  loss_rpn_cls: 0.003406  loss_rpn_loc: 0.006904  time: 0.2704  data_time: 0.0128  lr: 0.00051948  max_mem: 2895M
[05/09 00:49:52 d2.utils.events]:  eta: 0:03:09  iter: 279  total_loss: 0.5366  loss_cls: 0.2182  loss_box_reg: 0.2895  loss_rpn_cls: 0.004898  loss_rpn_loc: 0.00624  time: 0.2697  data_time: 0.0141  lr: 0.00055944  max_mem: 2895M
[05/09 00:49:58 d2.utils.events]:  eta: 0:03:04  iter: 299  total_loss: 0.3997  loss_cls: 0.1956  loss_box_reg: 0.1964  loss_rpn_cls: 0.0007925  loss_rpn_loc: 0.006658  time: 0.2694  data_time: 0.0117  lr: 0.0005994  max_mem: 2895M
[05/09 00:50:03 d2.utils.events]:  eta: 0:02:58  iter: 319  total_loss: 0.5477  loss_cls: 0.2752  loss_box_reg: 0.2546  loss_rpn_cls: 0.002897  loss_rpn_loc: 0.005303  time: 0.2688  data_time: 0.0124  lr: 0.00063936  max_mem: 2895M
[05/09 00:50:08 d2.utils.events]:  eta: 0:02:54  iter: 339  total_loss: 0.4718  loss_cls: 0.2473  loss_box_reg: 0.2481  loss_rpn_cls: 0.003662  loss_rpn_loc: 0.004988  time: 0.2694  data_time: 0.0141  lr: 0.00067932  max_mem: 2895M
[05/09 00:50:14 d2.utils.events]:  eta: 0:02:49  iter: 359  total_loss: 0.4421  loss_cls: 0.208  loss_box_reg: 0.2091  loss_rpn_cls: 0.001212  loss_rpn_loc: 0.004857  time: 0.2693  data_time: 0.0143  lr: 0.00071928  max_mem: 2895M
[05/09 00:50:19 d2.utils.events]:  eta: 0:02:43  iter: 379  total_loss: 0.5215  loss_cls: 0.2466  loss_box_reg: 0.2727  loss_rpn_cls: 0.005805  loss_rpn_loc: 0.006626  time: 0.2691  data_time: 0.0108  lr: 0.00075924  max_mem: 2895M
[05/09 00:50:24 d2.utils.events]:  eta: 0:02:39  iter: 399  total_loss: 0.4769  loss_cls: 0.2007  loss_box_reg: 0.257  loss_rpn_cls: 0.001677  loss_rpn_loc: 0.005929  time: 0.2689  data_time: 0.0095  lr: 0.0007992  max_mem: 2895M
[05/09 00:50:30 d2.utils.events]:  eta: 0:02:34  iter: 419  total_loss: 0.4454  loss_cls: 0.2009  loss_box_reg: 0.2315  loss_rpn_cls: 0.001209  loss_rpn_loc: 0.006166  time: 0.2692  data_time: 0.0124  lr: 0.00083916  max_mem: 2895M
[05/09 00:50:35 d2.utils.events]:  eta: 0:02:28  iter: 439  total_loss: 0.4667  loss_cls: 0.1965  loss_box_reg: 0.2467  loss_rpn_cls: 0.00139  loss_rpn_loc: 0.006625  time: 0.2689  data_time: 0.0124  lr: 0.00087912  max_mem: 2895M
[05/09 00:50:41 d2.utils.events]:  eta: 0:02:23  iter: 459  total_loss: 0.4424  loss_cls: 0.2092  loss_box_reg: 0.2734  loss_rpn_cls: 0.0009896  loss_rpn_loc: 0.005665  time: 0.2689  data_time: 0.0115  lr: 0.00091908  max_mem: 2895M
[05/09 00:50:46 d2.utils.events]:  eta: 0:02:17  iter: 479  total_loss: 0.4396  loss_cls: 0.2091  loss_box_reg: 0.2079  loss_rpn_cls: 0.002499  loss_rpn_loc: 0.007423  time: 0.2685  data_time: 0.0109  lr: 0.00095904  max_mem: 2895M
[05/09 00:50:51 d2.utils.events]:  eta: 0:02:12  iter: 499  total_loss: 0.4761  loss_cls: 0.215  loss_box_reg: 0.2123  loss_rpn_cls: 0.000724  loss_rpn_loc: 0.00729  time: 0.2684  data_time: 0.0127  lr: 0.000999  max_mem: 2895M
[05/09 00:50:56 d2.utils.events]:  eta: 0:02:07  iter: 519  total_loss: 0.4595  loss_cls: 0.2071  loss_box_reg: 0.2559  loss_rpn_cls: 0.0009723  loss_rpn_loc: 0.006114  time: 0.2683  data_time: 0.0121  lr: 0.001039  max_mem: 2895M
[05/09 00:51:02 d2.utils.events]:  eta: 0:02:01  iter: 539  total_loss: 0.4348  loss_cls: 0.1874  loss_box_reg: 0.2136  loss_rpn_cls: 0.003308  loss_rpn_loc: 0.005711  time: 0.2681  data_time: 0.0160  lr: 0.0010789  max_mem: 2895M
[05/09 00:51:07 d2.utils.events]:  eta: 0:01:56  iter: 559  total_loss: 0.4391  loss_cls: 0.2033  loss_box_reg: 0.1957  loss_rpn_cls: 0.00218  loss_rpn_loc: 0.004848  time: 0.2682  data_time: 0.0107  lr: 0.0011189  max_mem: 2895M
[05/09 00:51:13 d2.utils.events]:  eta: 0:01:51  iter: 579  total_loss: 0.4409  loss_cls: 0.223  loss_box_reg: 0.2327  loss_rpn_cls: 0.0008134  loss_rpn_loc: 0.005215  time: 0.2681  data_time: 0.0130  lr: 0.0011588  max_mem: 2895M
[05/09 00:51:18 d2.utils.events]:  eta: 0:01:46  iter: 599  total_loss: 0.3866  loss_cls: 0.1817  loss_box_reg: 0.1796  loss_rpn_cls: 0.001697  loss_rpn_loc: 0.007755  time: 0.2682  data_time: 0.0109  lr: 0.0011988  max_mem: 2895M
[05/09 00:51:23 d2.utils.events]:  eta: 0:01:40  iter: 619  total_loss: 0.4026  loss_cls: 0.1602  loss_box_reg: 0.2436  loss_rpn_cls: 0.002327  loss_rpn_loc: 0.005093  time: 0.2682  data_time: 0.0144  lr: 0.0012388  max_mem: 2895M
[05/09 00:51:29 d2.utils.events]:  eta: 0:01:35  iter: 639  total_loss: 0.372  loss_cls: 0.1553  loss_box_reg: 0.1873  loss_rpn_cls: 0.003113  loss_rpn_loc: 0.00546  time: 0.2680  data_time: 0.0117  lr: 0.0012787  max_mem: 2895M
[05/09 00:51:34 d2.utils.events]:  eta: 0:01:30  iter: 659  total_loss: 0.4043  loss_cls: 0.2116  loss_box_reg: 0.2117  loss_rpn_cls: 0.0008684  loss_rpn_loc: 0.005729  time: 0.2680  data_time: 0.0121  lr: 0.0013187  max_mem: 2895M
[05/09 00:51:39 d2.utils.events]:  eta: 0:01:24  iter: 679  total_loss: 0.5002  loss_cls: 0.2232  loss_box_reg: 0.2555  loss_rpn_cls: 0.001154  loss_rpn_loc: 0.005423  time: 0.2681  data_time: 0.0100  lr: 0.0013586  max_mem: 2895M
[05/09 00:51:45 d2.utils.events]:  eta: 0:01:19  iter: 699  total_loss: 0.392  loss_cls: 0.1622  loss_box_reg: 0.2408  loss_rpn_cls: 0.0014  loss_rpn_loc: 0.005848  time: 0.2681  data_time: 0.0141  lr: 0.0013986  max_mem: 2895M
[05/09 00:51:50 d2.utils.events]:  eta: 0:01:14  iter: 719  total_loss: 0.4301  loss_cls: 0.2027  loss_box_reg: 0.2085  loss_rpn_cls: 0.001928  loss_rpn_loc: 0.006883  time: 0.2678  data_time: 0.0123  lr: 0.0014386  max_mem: 2895M
[05/09 00:51:55 d2.utils.events]:  eta: 0:01:08  iter: 739  total_loss: 0.4679  loss_cls: 0.228  loss_box_reg: 0.2122  loss_rpn_cls: 0.001007  loss_rpn_loc: 0.005277  time: 0.2680  data_time: 0.0140  lr: 0.0014785  max_mem: 2895M
[05/09 00:52:01 d2.utils.events]:  eta: 0:01:03  iter: 759  total_loss: 0.3661  loss_cls: 0.1746  loss_box_reg: 0.1698  loss_rpn_cls: 0.00128  loss_rpn_loc: 0.007702  time: 0.2679  data_time: 0.0115  lr: 0.0015185  max_mem: 2895M
[05/09 00:52:06 d2.utils.events]:  eta: 0:00:58  iter: 779  total_loss: 0.4373  loss_cls: 0.1957  loss_box_reg: 0.243  loss_rpn_cls: 0.001587  loss_rpn_loc: 0.004933  time: 0.2679  data_time: 0.0118  lr: 0.0015584  max_mem: 2895M
[05/09 00:52:11 d2.utils.events]:  eta: 0:00:52  iter: 799  total_loss: 0.4411  loss_cls: 0.2009  loss_box_reg: 0.2033  loss_rpn_cls: 0.0007105  loss_rpn_loc: 0.004162  time: 0.2678  data_time: 0.0112  lr: 0.0015984  max_mem: 2895M
[05/09 00:52:17 d2.utils.events]:  eta: 0:00:47  iter: 819  total_loss: 0.4095  loss_cls: 0.1617  loss_box_reg: 0.2326  loss_rpn_cls: 0.003482  loss_rpn_loc: 0.003971  time: 0.2676  data_time: 0.0136  lr: 0.0016384  max_mem: 2895M
[05/09 00:52:22 d2.utils.events]:  eta: 0:00:42  iter: 839  total_loss: 0.3875  loss_cls: 0.206  loss_box_reg: 0.1927  loss_rpn_cls: 0.001513  loss_rpn_loc: 0.004942  time: 0.2675  data_time: 0.0136  lr: 0.0016783  max_mem: 2895M
[05/09 00:52:27 d2.utils.events]:  eta: 0:00:37  iter: 859  total_loss: 0.3919  loss_cls: 0.1638  loss_box_reg: 0.2097  loss_rpn_cls: 0.0009756  loss_rpn_loc: 0.005418  time: 0.2675  data_time: 0.0126  lr: 0.0017183  max_mem: 2895M
[05/09 00:52:33 d2.utils.events]:  eta: 0:00:31  iter: 879  total_loss: 0.5211  loss_cls: 0.2122  loss_box_reg: 0.2564  loss_rpn_cls: 0.002488  loss_rpn_loc: 0.006843  time: 0.2675  data_time: 0.0119  lr: 0.0017582  max_mem: 2895M
[05/09 00:52:38 d2.utils.events]:  eta: 0:00:26  iter: 899  total_loss: 0.4934  loss_cls: 0.1712  loss_box_reg: 0.2678  loss_rpn_cls: 0.001097  loss_rpn_loc: 0.006459  time: 0.2676  data_time: 0.0116  lr: 0.0017982  max_mem: 2895M
[05/09 00:52:43 d2.utils.events]:  eta: 0:00:21  iter: 919  total_loss: 0.4764  loss_cls: 0.1865  loss_box_reg: 0.2495  loss_rpn_cls: 0.001832  loss_rpn_loc: 0.006316  time: 0.2677  data_time: 0.0123  lr: 0.0018382  max_mem: 2895M
[05/09 00:52:49 d2.utils.events]:  eta: 0:00:15  iter: 939  total_loss: 0.5045  loss_cls: 0.2194  loss_box_reg: 0.2934  loss_rpn_cls: 0.002098  loss_rpn_loc: 0.006287  time: 0.2675  data_time: 0.0109  lr: 0.0018781  max_mem: 2895M
[05/09 00:52:54 d2.utils.events]:  eta: 0:00:10  iter: 959  total_loss: 0.4933  loss_cls: 0.213  loss_box_reg: 0.2692  loss_rpn_cls: 0.002477  loss_rpn_loc: 0.007896  time: 0.2674  data_time: 0.0132  lr: 0.0019181  max_mem: 2895M
[05/09 00:52:59 d2.utils.events]:  eta: 0:00:05  iter: 979  total_loss: 0.4543  loss_cls: 0.2021  loss_box_reg: 0.2715  loss_rpn_cls: 0.001956  loss_rpn_loc: 0.005839  time: 0.2673  data_time: 0.0134  lr: 0.001958  max_mem: 2895M
[05/09 00:53:06 d2.utils.events]:  eta: 0:00:00  iter: 999  total_loss: 0.4072  loss_cls: 0.1676  loss_box_reg: 0.2348  loss_rpn_cls: 0.001137  loss_rpn_loc: 0.005125  time: 0.2675  data_time: 0.0122  lr: 0.001998  max_mem: 2895M
[05/09 00:53:06 d2.engine.hooks]: Overall training speed: 998 iterations in 0:04:26 (0.2675 s / it)
[05/09 00:53:06 d2.engine.hooks]: Total training time: 0:04:29 (0:00:02 on hooks)

cfg.MODEL.WEIGHTS = os.path.join(cfg.OUTPUT_DIR, "model_final.pth")
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5   # set the testing threshold for this model
cfg.DATASETS.TEST = ("rel_actions_val3", )
predictor = DefaultPredictor(cfg)

[05/09 00:53:21 d2.checkpoint.c2_model_loading]: Following weights matched with model:
| Names in Model                                  | Names in Checkpoint                                                                                  | Shapes                                          |
|:------------------------------------------------|:-----------------------------------------------------------------------------------------------------|:------------------------------------------------|
| backbone.bottom_up.res2.0.conv1.*               | backbone.bottom_up.res2.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,1,1)             |
| backbone.bottom_up.res2.0.conv2.*               | backbone.bottom_up.res2.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.0.conv3.*               | backbone.bottom_up.res2.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.0.shortcut.*            | backbone.bottom_up.res2.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.1.conv1.*               | backbone.bottom_up.res2.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.1.conv2.*               | backbone.bottom_up.res2.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.1.conv3.*               | backbone.bottom_up.res2.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.2.conv1.*               | backbone.bottom_up.res2.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.2.conv2.*               | backbone.bottom_up.res2.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.2.conv3.*               | backbone.bottom_up.res2.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res3.0.conv1.*               | backbone.bottom_up.res3.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,256,1,1)       |
| backbone.bottom_up.res3.0.conv2.*               | backbone.bottom_up.res3.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.0.conv3.*               | backbone.bottom_up.res3.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.0.shortcut.*            | backbone.bottom_up.res3.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (512,) (512,) (512,) (512,) (512,256,1,1)       |
| backbone.bottom_up.res3.1.conv1.*               | backbone.bottom_up.res3.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.1.conv2.*               | backbone.bottom_up.res3.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.1.conv3.*               | backbone.bottom_up.res3.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.2.conv1.*               | backbone.bottom_up.res3.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.2.conv2.*               | backbone.bottom_up.res3.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.2.conv3.*               | backbone.bottom_up.res3.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.3.conv1.*               | backbone.bottom_up.res3.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.3.conv2.*               | backbone.bottom_up.res3.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.3.conv3.*               | backbone.bottom_up.res3.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res4.0.conv1.*               | backbone.bottom_up.res4.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,512,1,1)       |
| backbone.bottom_up.res4.0.conv2.*               | backbone.bottom_up.res4.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.0.conv3.*               | backbone.bottom_up.res4.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.0.shortcut.*            | backbone.bottom_up.res4.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (1024,) (1024,) (1024,) (1024,) (1024,512,1,1)  |
| backbone.bottom_up.res4.1.conv1.*               | backbone.bottom_up.res4.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.1.conv2.*               | backbone.bottom_up.res4.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.1.conv3.*               | backbone.bottom_up.res4.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.2.conv1.*               | backbone.bottom_up.res4.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.2.conv2.*               | backbone.bottom_up.res4.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.2.conv3.*               | backbone.bottom_up.res4.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.3.conv1.*               | backbone.bottom_up.res4.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.3.conv2.*               | backbone.bottom_up.res4.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.3.conv3.*               | backbone.bottom_up.res4.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.4.conv1.*               | backbone.bottom_up.res4.4.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.4.conv2.*               | backbone.bottom_up.res4.4.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.4.conv3.*               | backbone.bottom_up.res4.4.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.5.conv1.*               | backbone.bottom_up.res4.5.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.5.conv2.*               | backbone.bottom_up.res4.5.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.5.conv3.*               | backbone.bottom_up.res4.5.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res5.0.conv1.*               | backbone.bottom_up.res5.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,1024,1,1)      |
| backbone.bottom_up.res5.0.conv2.*               | backbone.bottom_up.res5.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.0.conv3.*               | backbone.bottom_up.res5.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.0.shortcut.*            | backbone.bottom_up.res5.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (2048,) (2048,) (2048,) (2048,) (2048,1024,1,1) |
| backbone.bottom_up.res5.1.conv1.*               | backbone.bottom_up.res5.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.1.conv2.*               | backbone.bottom_up.res5.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.1.conv3.*               | backbone.bottom_up.res5.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.2.conv1.*               | backbone.bottom_up.res5.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.2.conv2.*               | backbone.bottom_up.res5.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.2.conv3.*               | backbone.bottom_up.res5.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.stem.conv1.*                 | backbone.bottom_up.stem.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}      | (64,) (64,) (64,) (64,) (64,3,7,7)              |
| backbone.fpn_lateral2.*                         | backbone.fpn_lateral2.{bias,weight}                                                                  | (256,) (256,256,1,1)                            |
| backbone.fpn_lateral3.*                         | backbone.fpn_lateral3.{bias,weight}                                                                  | (256,) (256,512,1,1)                            |
| backbone.fpn_lateral4.*                         | backbone.fpn_lateral4.{bias,weight}                                                                  | (256,) (256,1024,1,1)                           |
| backbone.fpn_lateral5.*                         | backbone.fpn_lateral5.{bias,weight}                                                                  | (256,) (256,2048,1,1)                           |
| backbone.fpn_output2.*                          | backbone.fpn_output2.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output3.*                          | backbone.fpn_output3.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output4.*                          | backbone.fpn_output4.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output5.*                          | backbone.fpn_output5.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.anchor_deltas.*     | proposal_generator.rpn_head.anchor_deltas.{bias,weight}                                              | (12,) (12,256,1,1)                              |
| proposal_generator.rpn_head.conv.*              | proposal_generator.rpn_head.conv.{bias,weight}                                                       | (256,) (256,256,3,3)                            |
| proposal_generator.rpn_head.objectness_logits.* | proposal_generator.rpn_head.objectness_logits.{bias,weight}                                          | (3,) (3,256,1,1)                                |
| roi_heads.box_head.fc1.*                        | roi_heads.box_head.fc1.{bias,weight}                                                                 | (1024,) (1024,12544)                            |
| roi_heads.box_head.fc2.*                        | roi_heads.box_head.fc2.{bias,weight}                                                                 | (1024,) (1024,1024)                             |
| roi_heads.box_predictor.bbox_pred.*             | roi_heads.box_predictor.bbox_pred.{bias,weight}                                                      | (28,) (28,1024)                                 |
| roi_heads.box_predictor.cls_score.*             | roi_heads.box_predictor.cls_score.{bias,weight}                                                      | (8,) (8,1024)                                   |

from detectron2.evaluation import COCOEvaluator, inference_on_dataset
from detectron2.data import build_detection_test_loader
from detectron2.engine import DefaultTrainer

evaluator = COCOEvaluator("rel_actions_val3", cfg, False, output_dir="./output_evals/")
val_loader = build_detection_test_loader(cfg, "rel_actions_val3")
inference_on_dataset(trainer.model, val_loader, evaluator)

WARNING [05/09 00:53:28 d2.evaluation.coco_evaluation]: COCO Evaluator instantiated using config, this is deprecated behavior. Please pass in explicit arguments instead.
[05/09 00:53:28 d2.data.datasets.coco]: Loaded 2215 images in COCO format from ./data/instances_vcoco_actions_val_2014.json
[05/09 00:53:29 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in inference: [ResizeShortestEdge(short_edge_length=(800, 800), max_size=1333, sample_style='choice')]
[05/09 00:53:29 d2.data.common]: Serializing 2215 elements to byte tensors and concatenating them all ...
[05/09 00:53:29 d2.data.common]: Serialized dataset takes 5.66 MiB
[05/09 00:53:29 d2.evaluation.evaluator]: Start inference on 2215 batches
[05/09 00:53:30 d2.evaluation.evaluator]: Inference done 11/2215. Dataloading: 0.0016 s/iter. Inference: 0.0630 s/iter. Eval: 0.0003 s/iter. Total: 0.0649 s/iter. ETA=0:02:22
[05/09 00:53:35 d2.evaluation.evaluator]: Inference done 84/2215. Dataloading: 0.0022 s/iter. Inference: 0.0662 s/iter. Eval: 0.0003 s/iter. Total: 0.0688 s/iter. ETA=0:02:26
[05/09 00:53:40 d2.evaluation.evaluator]: Inference done 159/2215. Dataloading: 0.0023 s/iter. Inference: 0.0654 s/iter. Eval: 0.0003 s/iter. Total: 0.0681 s/iter. ETA=0:02:19
[05/09 00:53:45 d2.evaluation.evaluator]: Inference done 232/2215. Dataloading: 0.0023 s/iter. Inference: 0.0656 s/iter. Eval: 0.0003 s/iter. Total: 0.0683 s/iter. ETA=0:02:15
[05/09 00:53:50 d2.evaluation.evaluator]: Inference done 305/2215. Dataloading: 0.0023 s/iter. Inference: 0.0658 s/iter. Eval: 0.0003 s/iter. Total: 0.0685 s/iter. ETA=0:02:10
[05/09 00:53:55 d2.evaluation.evaluator]: Inference done 378/2215. Dataloading: 0.0023 s/iter. Inference: 0.0659 s/iter. Eval: 0.0003 s/iter. Total: 0.0686 s/iter. ETA=0:02:05
[05/09 00:54:00 d2.evaluation.evaluator]: Inference done 452/2215. Dataloading: 0.0023 s/iter. Inference: 0.0659 s/iter. Eval: 0.0003 s/iter. Total: 0.0685 s/iter. ETA=0:02:00
[05/09 00:54:05 d2.evaluation.evaluator]: Inference done 526/2215. Dataloading: 0.0023 s/iter. Inference: 0.0658 s/iter. Eval: 0.0003 s/iter. Total: 0.0684 s/iter. ETA=0:01:55
[05/09 00:54:10 d2.evaluation.evaluator]: Inference done 599/2215. Dataloading: 0.0023 s/iter. Inference: 0.0658 s/iter. Eval: 0.0003 s/iter. Total: 0.0685 s/iter. ETA=0:01:50
[05/09 00:54:15 d2.evaluation.evaluator]: Inference done 671/2215. Dataloading: 0.0023 s/iter. Inference: 0.0659 s/iter. Eval: 0.0003 s/iter. Total: 0.0686 s/iter. ETA=0:01:45
[05/09 00:54:20 d2.evaluation.evaluator]: Inference done 743/2215. Dataloading: 0.0023 s/iter. Inference: 0.0660 s/iter. Eval: 0.0003 s/iter. Total: 0.0687 s/iter. ETA=0:01:41
[05/09 00:54:25 d2.evaluation.evaluator]: Inference done 815/2215. Dataloading: 0.0023 s/iter. Inference: 0.0661 s/iter. Eval: 0.0003 s/iter. Total: 0.0688 s/iter. ETA=0:01:36
[05/09 00:54:30 d2.evaluation.evaluator]: Inference done 888/2215. Dataloading: 0.0023 s/iter. Inference: 0.0662 s/iter. Eval: 0.0003 s/iter. Total: 0.0688 s/iter. ETA=0:01:31
[05/09 00:54:35 d2.evaluation.evaluator]: Inference done 959/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:01:26
[05/09 00:54:40 d2.evaluation.evaluator]: Inference done 1032/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:01:21
[05/09 00:54:45 d2.evaluation.evaluator]: Inference done 1105/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:01:16
[05/09 00:54:50 d2.evaluation.evaluator]: Inference done 1178/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:01:11
[05/09 00:54:55 d2.evaluation.evaluator]: Inference done 1250/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:01:06
[05/09 00:55:00 d2.evaluation.evaluator]: Inference done 1322/2215. Dataloading: 0.0023 s/iter. Inference: 0.0664 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:01:01
[05/09 00:55:05 d2.evaluation.evaluator]: Inference done 1395/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:56
[05/09 00:55:10 d2.evaluation.evaluator]: Inference done 1468/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:51
[05/09 00:55:15 d2.evaluation.evaluator]: Inference done 1541/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:00:46
[05/09 00:55:21 d2.evaluation.evaluator]: Inference done 1613/2215. Dataloading: 0.0023 s/iter. Inference: 0.0664 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:00:41
[05/09 00:55:26 d2.evaluation.evaluator]: Inference done 1686/2215. Dataloading: 0.0023 s/iter. Inference: 0.0664 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:00:36
[05/09 00:55:31 d2.evaluation.evaluator]: Inference done 1759/2215. Dataloading: 0.0023 s/iter. Inference: 0.0664 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:00:31
[05/09 00:55:36 d2.evaluation.evaluator]: Inference done 1833/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0691 s/iter. ETA=0:00:26
[05/09 00:55:41 d2.evaluation.evaluator]: Inference done 1907/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:21
[05/09 00:55:46 d2.evaluation.evaluator]: Inference done 1979/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:16
[05/09 00:55:51 d2.evaluation.evaluator]: Inference done 2053/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:11
[05/09 00:55:56 d2.evaluation.evaluator]: Inference done 2126/2215. Dataloading: 0.0023 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:06
[05/09 00:56:01 d2.evaluation.evaluator]: Inference done 2199/2215. Dataloading: 0.0024 s/iter. Inference: 0.0663 s/iter. Eval: 0.0003 s/iter. Total: 0.0690 s/iter. ETA=0:00:01
[05/09 00:56:02 d2.evaluation.evaluator]: Total inference time: 0:02:32.564105 (0.069034 s / iter per device, on 1 devices)
[05/09 00:56:02 d2.evaluation.evaluator]: Total inference pure compute time: 0:02:26 (0.066264 s / iter per device, on 1 devices)
[05/09 00:56:02 d2.evaluation.coco_evaluation]: Preparing results for COCO format ...
[05/09 00:56:02 d2.evaluation.coco_evaluation]: Saving results to ./output_evals/coco_instances_results.json
[05/09 00:56:02 d2.evaluation.coco_evaluation]: Evaluating predictions with unofficial COCO API...
Loading and preparing results...
DONE (t=0.05s)
creating index...
index created!
[05/09 00:56:02 d2.evaluation.fast_eval_api]: Evaluate annotation type *bbox*
[05/09 00:56:03 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 0.72 seconds.
[05/09 00:56:03 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 00:56:03 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.15 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.108
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.170
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.124
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.152
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.114
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.115
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.206
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.206
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.500
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.235
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.207
[05/09 00:56:03 d2.evaluation.coco_evaluation]: Evaluation results for bbox: 
|   AP   |  AP50  |  AP75  |  APs   |  APm   |  APl   |
|:------:|:------:|:------:|:------:|:------:|:------:|
| 10.818 | 16.965 | 12.392 | 50.000 | 15.232 | 11.430 |
[05/09 00:56:03 d2.evaluation.coco_evaluation]: Per-category bbox AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 39.873 | carry      | 0.000 | point      | 0.000  |
| eat           | 0.423  | drink      | 0.000 | stand      | 35.430 |
| talk_on_phone | 0.000  |            |       |            |        |

OrderedDict([('bbox',
              {'AP': 10.818063676259921,
               'AP-carry': 0.0,
               'AP-drink': 0.0,
               'AP-eat': 0.4225658325436689,
               'AP-hold': 39.87349445213421,
               'AP-point': 0.0,
               'AP-stand': 35.43038544914159,
               'AP-talk_on_phone': 0.0,
               'AP50': 16.964607948315617,
               'AP75': 12.391597443236359,
               'APl': 11.429539806129627,
               'APm': 15.231991371583723,
               'APs': 50.0})])

Train on Retinent Model 50

from detectron2.engine import DefaultTrainer
from detectron2.config import get_cfg
import os

from detectron2 import model_zoo

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file("COCO-Detection/retinanet_R_101_FPN_3x.yaml"))
cfg.DATASETS.TRAIN = ("rel_actions1",)
cfg.DATASETS.TEST = ()   # no metrics implemented for this dataset
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url("COCO-Detection/retinanet_R_101_FPN_3x.yaml")  # Let training initialize from model zoo
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.SOLVER.BASE_LR = 0.002
cfg.SOLVER.MAX_ITER = 1000    
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 128   # faster, and good enough for this toy dataset
# cfg.MODEL.ROI_HEADS.NUM_CLASSES = len(relevant_classes)  # all relevant classes
cfg.MODEL.RETINANET.NUM_CLASSES = len(relevant_classes)
cfg.OUTPUT_DIR = "./outputs/action_only_detector"

os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
trainer = DefaultTrainer(cfg)
trainer.resume_or_load(resume=False)
trainer.train()

Loading config /content/drive/.shortcut-targets-by-id/1Eg2dTJ9lloHImzZiTfOZ7_jOP1sTh--c/6.869 Project/v-coco/detectron2_repo/detectron2/model_zoo/configs/COCO-Detection/../Base-RetinaNet.yaml with yaml.unsafe_load. Your machine may be at risk if the file contains malicious content.

[05/09 01:02:54 d2.engine.defaults]: Model:
RetinaNet(
  (backbone): FPN(
    (fpn_lateral3): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral4): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output4): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (fpn_lateral5): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
    (fpn_output5): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (top_block): LastLevelP6P7(
      (p6): Conv2d(2048, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1))
      (p7): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1))
    )
    (bottom_up): ResNet(
      (stem): BasicStem(
        (conv1): Conv2d(
          3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False
          (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
        )
      )
      (res2): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv1): Conv2d(
            64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv2): Conv2d(
            64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
          )
          (conv3): Conv2d(
            64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
        )
      )
      (res3): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv1): Conv2d(
            256, 128, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv2): Conv2d(
            128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
          )
          (conv3): Conv2d(
            128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
        )
      )
      (res4): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
          (conv1): Conv2d(
            512, 256, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (3): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (4): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (5): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (6): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (7): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (8): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (9): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (10): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (11): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (12): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (13): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (14): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (15): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (16): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (17): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (18): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (19): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (20): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (21): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
        (22): BottleneckBlock(
          (conv1): Conv2d(
            1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv2): Conv2d(
            256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
          )
          (conv3): Conv2d(
            256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
          )
        )
      )
      (res5): Sequential(
        (0): BottleneckBlock(
          (shortcut): Conv2d(
            1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
          (conv1): Conv2d(
            1024, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (1): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
        (2): BottleneckBlock(
          (conv1): Conv2d(
            2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv2): Conv2d(
            512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
          )
          (conv3): Conv2d(
            512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
            (norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
          )
        )
      )
    )
  )
  (head): RetinaNetHead(
    (cls_subnet): Sequential(
      (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (1): ReLU()
      (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (3): ReLU()
      (4): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (5): ReLU()
      (6): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (7): ReLU()
    )
    (bbox_subnet): Sequential(
      (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (1): ReLU()
      (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (3): ReLU()
      (4): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (5): ReLU()
      (6): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (7): ReLU()
    )
    (cls_score): Conv2d(256, 63, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (bbox_pred): Conv2d(256, 36, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  )
  (anchor_generator): DefaultAnchorGenerator(
    (cell_anchors): BufferList()
  )
)
[05/09 01:02:55 d2.data.datasets.coco]: Loaded 1951 images in COCO format from ./data/instances_vcoco_actions_train.json
[05/09 01:02:55 d2.data.build]: Removed 0 images with no usable annotations. 1951 images left.
[05/09 01:02:55 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in training: [ResizeShortestEdge(short_edge_length=(640, 672, 704, 736, 768, 800), max_size=1333, sample_style='choice'), RandomFlip()]
[05/09 01:02:55 d2.data.build]: Using training sampler TrainingSampler
[05/09 01:02:55 d2.data.common]: Serializing 1951 elements to byte tensors and concatenating them all ...
[05/09 01:02:55 d2.data.common]: Serialized dataset takes 4.90 MiB
WARNING [05/09 01:02:55 d2.solver.build]: SOLVER.STEPS contains values larger than SOLVER.MAX_ITER. These values will be ignored.

model_final_971ab9.pkl: 228MB [00:17, 12.8MB/s]                           
Skip loading parameter 'head.cls_score.weight' to the model due to incompatible shapes: (720, 256, 3, 3) in the checkpoint but (63, 256, 3, 3) in the model! You might want to double check if this is expected.
Skip loading parameter 'head.cls_score.bias' to the model due to incompatible shapes: (720,) in the checkpoint but (63,) in the model! You might want to double check if this is expected.
Some model parameters or buffers are not found in the checkpoint:
head.cls_score.{bias, weight}
The checkpoint state_dict contains keys that are not used by the model:
  pixel_mean
  pixel_std

[05/09 01:03:13 d2.engine.train_loop]: Starting training from iteration 0
[05/09 01:03:27 d2.utils.events]:  eta: 0:06:35  iter: 19  total_loss: 1.291  loss_cls: 1.149  loss_box_reg: 0.09464  time: 0.7037  data_time: 0.0223  lr: 3.9962e-05  max_mem: 3138M
[05/09 01:03:34 d2.utils.events]:  eta: 0:05:57  iter: 39  total_loss: 1.104  loss_cls: 0.9954  loss_box_reg: 0.0798  time: 0.5247  data_time: 0.0123  lr: 7.9922e-05  max_mem: 3138M
[05/09 01:03:41 d2.utils.events]:  eta: 0:05:41  iter: 59  total_loss: 0.9339  loss_cls: 0.8588  loss_box_reg: 0.07811  time: 0.4640  data_time: 0.0104  lr: 0.00011988  max_mem: 3138M
[05/09 01:03:48 d2.utils.events]:  eta: 0:05:34  iter: 79  total_loss: 0.8149  loss_cls: 0.7471  loss_box_reg: 0.08631  time: 0.4387  data_time: 0.0119  lr: 0.00015984  max_mem: 3138M
[05/09 01:03:56 d2.utils.events]:  eta: 0:05:27  iter: 99  total_loss: 0.5795  loss_cls: 0.4521  loss_box_reg: 0.1185  time: 0.4256  data_time: 0.0115  lr: 0.0001998  max_mem: 3138M
[05/09 01:04:04 d2.utils.events]:  eta: 0:05:20  iter: 119  total_loss: 0.4047  loss_cls: 0.3367  loss_box_reg: 0.0666  time: 0.4191  data_time: 0.0104  lr: 0.00023976  max_mem: 3194M
[05/09 01:04:11 d2.utils.events]:  eta: 0:05:13  iter: 139  total_loss: 0.4907  loss_cls: 0.3991  loss_box_reg: 0.08291  time: 0.4126  data_time: 0.0117  lr: 0.00027972  max_mem: 3194M
[05/09 01:04:19 d2.utils.events]:  eta: 0:05:05  iter: 159  total_loss: 0.4396  loss_cls: 0.3368  loss_box_reg: 0.09362  time: 0.4077  data_time: 0.0156  lr: 0.00031968  max_mem: 3194M
[05/09 01:04:27 d2.utils.events]:  eta: 0:04:59  iter: 179  total_loss: 0.407  loss_cls: 0.3104  loss_box_reg: 0.08639  time: 0.4065  data_time: 0.0112  lr: 0.00035964  max_mem: 3196M
[05/09 01:04:35 d2.utils.events]:  eta: 0:04:55  iter: 199  total_loss: 0.4064  loss_cls: 0.2811  loss_box_reg: 0.1042  time: 0.4073  data_time: 0.0196  lr: 0.0003996  max_mem: 3196M
[05/09 01:04:42 d2.utils.events]:  eta: 0:04:47  iter: 219  total_loss: 0.4155  loss_cls: 0.3076  loss_box_reg: 0.1106  time: 0.4038  data_time: 0.0110  lr: 0.00043956  max_mem: 3196M
[05/09 01:04:50 d2.utils.events]:  eta: 0:04:39  iter: 239  total_loss: 0.2786  loss_cls: 0.201  loss_box_reg: 0.07397  time: 0.4000  data_time: 0.0100  lr: 0.00047952  max_mem: 3196M
[05/09 01:04:57 d2.utils.events]:  eta: 0:04:32  iter: 259  total_loss: 0.3907  loss_cls: 0.2901  loss_box_reg: 0.07736  time: 0.3987  data_time: 0.0121  lr: 0.00051948  max_mem: 3196M
[05/09 01:05:05 d2.utils.events]:  eta: 0:04:25  iter: 279  total_loss: 0.3119  loss_cls: 0.2405  loss_box_reg: 0.07196  time: 0.3970  data_time: 0.0118  lr: 0.00055944  max_mem: 3196M
[05/09 01:05:12 d2.utils.events]:  eta: 0:04:17  iter: 299  total_loss: 0.3761  loss_cls: 0.2507  loss_box_reg: 0.09403  time: 0.3945  data_time: 0.0122  lr: 0.0005994  max_mem: 3196M
[05/09 01:05:19 d2.utils.events]:  eta: 0:04:09  iter: 319  total_loss: 0.3592  loss_cls: 0.2351  loss_box_reg: 0.09504  time: 0.3922  data_time: 0.0115  lr: 0.00063936  max_mem: 3196M
[05/09 01:05:26 d2.utils.events]:  eta: 0:04:01  iter: 339  total_loss: 0.4348  loss_cls: 0.2863  loss_box_reg: 0.1556  time: 0.3908  data_time: 0.0143  lr: 0.00067932  max_mem: 3196M
[05/09 01:05:34 d2.utils.events]:  eta: 0:03:54  iter: 359  total_loss: 0.3023  loss_cls: 0.2222  loss_box_reg: 0.07102  time: 0.3898  data_time: 0.0156  lr: 0.00071928  max_mem: 3196M
[05/09 01:05:41 d2.utils.events]:  eta: 0:03:46  iter: 379  total_loss: 0.434  loss_cls: 0.285  loss_box_reg: 0.1336  time: 0.3881  data_time: 0.0127  lr: 0.00075924  max_mem: 3196M
[05/09 01:05:48 d2.utils.events]:  eta: 0:03:39  iter: 399  total_loss: 0.4044  loss_cls: 0.3018  loss_box_reg: 0.1027  time: 0.3867  data_time: 0.0126  lr: 0.0007992  max_mem: 3196M
[05/09 01:05:56 d2.utils.events]:  eta: 0:03:31  iter: 419  total_loss: 0.4142  loss_cls: 0.2887  loss_box_reg: 0.1244  time: 0.3858  data_time: 0.0161  lr: 0.00083916  max_mem: 3196M
[05/09 01:06:03 d2.utils.events]:  eta: 0:03:24  iter: 439  total_loss: 0.4155  loss_cls: 0.2626  loss_box_reg: 0.1158  time: 0.3845  data_time: 0.0118  lr: 0.00087912  max_mem: 3196M
[05/09 01:06:10 d2.utils.events]:  eta: 0:03:16  iter: 459  total_loss: 0.346  loss_cls: 0.2178  loss_box_reg: 0.1042  time: 0.3837  data_time: 0.0130  lr: 0.00091908  max_mem: 3196M
[05/09 01:06:18 d2.utils.events]:  eta: 0:03:09  iter: 479  total_loss: 0.4098  loss_cls: 0.2707  loss_box_reg: 0.1092  time: 0.3833  data_time: 0.0130  lr: 0.00095904  max_mem: 3196M
[05/09 01:06:25 d2.utils.events]:  eta: 0:03:02  iter: 499  total_loss: 0.4552  loss_cls: 0.2975  loss_box_reg: 0.1461  time: 0.3831  data_time: 0.0134  lr: 0.000999  max_mem: 3196M
[05/09 01:06:33 d2.utils.events]:  eta: 0:02:55  iter: 519  total_loss: 0.4332  loss_cls: 0.279  loss_box_reg: 0.1033  time: 0.3825  data_time: 0.0125  lr: 0.001039  max_mem: 3196M
[05/09 01:06:40 d2.utils.events]:  eta: 0:02:47  iter: 539  total_loss: 0.3853  loss_cls: 0.2663  loss_box_reg: 0.1248  time: 0.3823  data_time: 0.0123  lr: 0.0010789  max_mem: 3196M
[05/09 01:06:48 d2.utils.events]:  eta: 0:02:40  iter: 559  total_loss: 0.3877  loss_cls: 0.2317  loss_box_reg: 0.1276  time: 0.3819  data_time: 0.0112  lr: 0.0011189  max_mem: 3196M
[05/09 01:06:55 d2.utils.events]:  eta: 0:02:33  iter: 579  total_loss: 0.3368  loss_cls: 0.2089  loss_box_reg: 0.1093  time: 0.3811  data_time: 0.0120  lr: 0.0011588  max_mem: 3196M
[05/09 01:07:02 d2.utils.events]:  eta: 0:02:25  iter: 599  total_loss: 0.3815  loss_cls: 0.2697  loss_box_reg: 0.1116  time: 0.3806  data_time: 0.0116  lr: 0.0011988  max_mem: 3196M
[05/09 01:07:09 d2.utils.events]:  eta: 0:02:18  iter: 619  total_loss: 0.3684  loss_cls: 0.2387  loss_box_reg: 0.09971  time: 0.3800  data_time: 0.0132  lr: 0.0012388  max_mem: 3196M
[05/09 01:07:17 d2.utils.events]:  eta: 0:02:10  iter: 639  total_loss: 0.4873  loss_cls: 0.3165  loss_box_reg: 0.1119  time: 0.3795  data_time: 0.0110  lr: 0.0012787  max_mem: 3196M
[05/09 01:07:24 d2.utils.events]:  eta: 0:02:03  iter: 659  total_loss: 0.3893  loss_cls: 0.2669  loss_box_reg: 0.1129  time: 0.3789  data_time: 0.0141  lr: 0.0013187  max_mem: 3196M
[05/09 01:07:31 d2.utils.events]:  eta: 0:01:56  iter: 679  total_loss: 0.485  loss_cls: 0.326  loss_box_reg: 0.1377  time: 0.3784  data_time: 0.0132  lr: 0.0013586  max_mem: 3196M
[05/09 01:07:39 d2.utils.events]:  eta: 0:01:49  iter: 699  total_loss: 0.4071  loss_cls: 0.2839  loss_box_reg: 0.14  time: 0.3782  data_time: 0.0130  lr: 0.0013986  max_mem: 3196M
[05/09 01:07:46 d2.utils.events]:  eta: 0:01:41  iter: 719  total_loss: 0.3932  loss_cls: 0.2767  loss_box_reg: 0.1146  time: 0.3776  data_time: 0.0135  lr: 0.0014386  max_mem: 3196M
[05/09 01:07:53 d2.utils.events]:  eta: 0:01:34  iter: 739  total_loss: 0.3812  loss_cls: 0.2876  loss_box_reg: 0.1232  time: 0.3775  data_time: 0.0152  lr: 0.0014785  max_mem: 3196M
[05/09 01:08:01 d2.utils.events]:  eta: 0:01:27  iter: 759  total_loss: 0.4658  loss_cls: 0.2737  loss_box_reg: 0.1709  time: 0.3774  data_time: 0.0113  lr: 0.0015185  max_mem: 3196M
[05/09 01:08:08 d2.utils.events]:  eta: 0:01:20  iter: 779  total_loss: 0.4578  loss_cls: 0.2988  loss_box_reg: 0.1496  time: 0.3772  data_time: 0.0114  lr: 0.0015584  max_mem: 3196M
[05/09 01:08:16 d2.utils.events]:  eta: 0:01:12  iter: 799  total_loss: 0.5417  loss_cls: 0.3343  loss_box_reg: 0.1399  time: 0.3771  data_time: 0.0133  lr: 0.0015984  max_mem: 3196M
[05/09 01:08:23 d2.utils.events]:  eta: 0:01:05  iter: 819  total_loss: 0.4969  loss_cls: 0.3531  loss_box_reg: 0.128  time: 0.3770  data_time: 0.0114  lr: 0.0016384  max_mem: 3196M
[05/09 01:08:30 d2.utils.events]:  eta: 0:00:58  iter: 839  total_loss: 0.4378  loss_cls: 0.289  loss_box_reg: 0.1266  time: 0.3767  data_time: 0.0113  lr: 0.0016783  max_mem: 3196M
[05/09 01:08:38 d2.utils.events]:  eta: 0:00:50  iter: 859  total_loss: 0.7647  loss_cls: 0.5698  loss_box_reg: 0.1361  time: 0.3765  data_time: 0.0127  lr: 0.0017183  max_mem: 3196M
[05/09 01:08:45 d2.utils.events]:  eta: 0:00:43  iter: 879  total_loss: 0.5889  loss_cls: 0.3976  loss_box_reg: 0.1799  time: 0.3765  data_time: 0.0145  lr: 0.0017582  max_mem: 3196M
[05/09 01:08:52 d2.utils.events]:  eta: 0:00:36  iter: 899  total_loss: 0.6933  loss_cls: 0.5052  loss_box_reg: 0.1531  time: 0.3760  data_time: 0.0105  lr: 0.0017982  max_mem: 3196M
[05/09 01:09:00 d2.utils.events]:  eta: 0:00:29  iter: 919  total_loss: 0.4321  loss_cls: 0.2709  loss_box_reg: 0.1463  time: 0.3758  data_time: 0.0117  lr: 0.0018382  max_mem: 3196M
[05/09 01:09:07 d2.utils.events]:  eta: 0:00:21  iter: 939  total_loss: 0.4459  loss_cls: 0.2775  loss_box_reg: 0.1288  time: 0.3757  data_time: 0.0113  lr: 0.0018781  max_mem: 3196M
[05/09 01:09:15 d2.utils.events]:  eta: 0:00:14  iter: 959  total_loss: 0.4141  loss_cls: 0.2734  loss_box_reg: 0.1423  time: 0.3757  data_time: 0.0124  lr: 0.0019181  max_mem: 3196M
[05/09 01:09:22 d2.utils.events]:  eta: 0:00:07  iter: 979  total_loss: 0.4112  loss_cls: 0.2543  loss_box_reg: 0.1521  time: 0.3754  data_time: 0.0140  lr: 0.001958  max_mem: 3196M
[05/09 01:09:31 d2.utils.events]:  eta: 0:00:00  iter: 999  total_loss: 0.4413  loss_cls: 0.2875  loss_box_reg: 0.1359  time: 0.3755  data_time: 0.0117  lr: 0.001998  max_mem: 3196M
[05/09 01:09:31 d2.engine.hooks]: Overall training speed: 998 iterations in 0:06:14 (0.3755 s / it)
[05/09 01:09:31 d2.engine.hooks]: Total training time: 0:06:17 (0:00:02 on hooks)

cfg.MODEL.WEIGHTS = os.path.join(cfg.OUTPUT_DIR, "model_final.pth")
# cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5   # set the testing threshold for this model
cfg.MODEL.RETINANET.SCORE_THRESH_TEST = 0.5
cfg.DATASETS.TEST = ("rel_actions_val3", )
predictor = DefaultPredictor(cfg)

[05/09 01:26:08 d2.checkpoint.c2_model_loading]: Following weights matched with model:
| Names in Model                       | Names in Checkpoint                                                                                  | Shapes                                          |
|:-------------------------------------|:-----------------------------------------------------------------------------------------------------|:------------------------------------------------|
| backbone.bottom_up.res2.0.conv1.*    | backbone.bottom_up.res2.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,1,1)             |
| backbone.bottom_up.res2.0.conv2.*    | backbone.bottom_up.res2.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.0.conv3.*    | backbone.bottom_up.res2.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.0.shortcut.* | backbone.bottom_up.res2.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.1.conv1.*    | backbone.bottom_up.res2.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.1.conv2.*    | backbone.bottom_up.res2.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.1.conv3.*    | backbone.bottom_up.res2.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res2.2.conv1.*    | backbone.bottom_up.res2.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,256,1,1)            |
| backbone.bottom_up.res2.2.conv2.*    | backbone.bottom_up.res2.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (64,) (64,) (64,) (64,) (64,64,3,3)             |
| backbone.bottom_up.res2.2.conv3.*    | backbone.bottom_up.res2.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,64,1,1)        |
| backbone.bottom_up.res3.0.conv1.*    | backbone.bottom_up.res3.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,256,1,1)       |
| backbone.bottom_up.res3.0.conv2.*    | backbone.bottom_up.res3.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.0.conv3.*    | backbone.bottom_up.res3.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.0.shortcut.* | backbone.bottom_up.res3.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (512,) (512,) (512,) (512,) (512,256,1,1)       |
| backbone.bottom_up.res3.1.conv1.*    | backbone.bottom_up.res3.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.1.conv2.*    | backbone.bottom_up.res3.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.1.conv3.*    | backbone.bottom_up.res3.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.2.conv1.*    | backbone.bottom_up.res3.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.2.conv2.*    | backbone.bottom_up.res3.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.2.conv3.*    | backbone.bottom_up.res3.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res3.3.conv1.*    | backbone.bottom_up.res3.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,512,1,1)       |
| backbone.bottom_up.res3.3.conv2.*    | backbone.bottom_up.res3.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (128,) (128,) (128,) (128,) (128,128,3,3)       |
| backbone.bottom_up.res3.3.conv3.*    | backbone.bottom_up.res3.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,128,1,1)       |
| backbone.bottom_up.res4.0.conv1.*    | backbone.bottom_up.res4.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,512,1,1)       |
| backbone.bottom_up.res4.0.conv2.*    | backbone.bottom_up.res4.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.0.conv3.*    | backbone.bottom_up.res4.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.0.shortcut.* | backbone.bottom_up.res4.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (1024,) (1024,) (1024,) (1024,) (1024,512,1,1)  |
| backbone.bottom_up.res4.1.conv1.*    | backbone.bottom_up.res4.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.1.conv2.*    | backbone.bottom_up.res4.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.1.conv3.*    | backbone.bottom_up.res4.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.10.conv1.*   | backbone.bottom_up.res4.10.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.10.conv2.*   | backbone.bottom_up.res4.10.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.10.conv3.*   | backbone.bottom_up.res4.10.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.11.conv1.*   | backbone.bottom_up.res4.11.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.11.conv2.*   | backbone.bottom_up.res4.11.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.11.conv3.*   | backbone.bottom_up.res4.11.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.12.conv1.*   | backbone.bottom_up.res4.12.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.12.conv2.*   | backbone.bottom_up.res4.12.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.12.conv3.*   | backbone.bottom_up.res4.12.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.13.conv1.*   | backbone.bottom_up.res4.13.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.13.conv2.*   | backbone.bottom_up.res4.13.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.13.conv3.*   | backbone.bottom_up.res4.13.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.14.conv1.*   | backbone.bottom_up.res4.14.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.14.conv2.*   | backbone.bottom_up.res4.14.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.14.conv3.*   | backbone.bottom_up.res4.14.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.15.conv1.*   | backbone.bottom_up.res4.15.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.15.conv2.*   | backbone.bottom_up.res4.15.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.15.conv3.*   | backbone.bottom_up.res4.15.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.16.conv1.*   | backbone.bottom_up.res4.16.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.16.conv2.*   | backbone.bottom_up.res4.16.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.16.conv3.*   | backbone.bottom_up.res4.16.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.17.conv1.*   | backbone.bottom_up.res4.17.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.17.conv2.*   | backbone.bottom_up.res4.17.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.17.conv3.*   | backbone.bottom_up.res4.17.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.18.conv1.*   | backbone.bottom_up.res4.18.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.18.conv2.*   | backbone.bottom_up.res4.18.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.18.conv3.*   | backbone.bottom_up.res4.18.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.19.conv1.*   | backbone.bottom_up.res4.19.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.19.conv2.*   | backbone.bottom_up.res4.19.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.19.conv3.*   | backbone.bottom_up.res4.19.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.2.conv1.*    | backbone.bottom_up.res4.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.2.conv2.*    | backbone.bottom_up.res4.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.2.conv3.*    | backbone.bottom_up.res4.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.20.conv1.*   | backbone.bottom_up.res4.20.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.20.conv2.*   | backbone.bottom_up.res4.20.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.20.conv3.*   | backbone.bottom_up.res4.20.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.21.conv1.*   | backbone.bottom_up.res4.21.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.21.conv2.*   | backbone.bottom_up.res4.21.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.21.conv3.*   | backbone.bottom_up.res4.21.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.22.conv1.*   | backbone.bottom_up.res4.22.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.22.conv2.*   | backbone.bottom_up.res4.22.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.22.conv3.*   | backbone.bottom_up.res4.22.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}   | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.3.conv1.*    | backbone.bottom_up.res4.3.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.3.conv2.*    | backbone.bottom_up.res4.3.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.3.conv3.*    | backbone.bottom_up.res4.3.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.4.conv1.*    | backbone.bottom_up.res4.4.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.4.conv2.*    | backbone.bottom_up.res4.4.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.4.conv3.*    | backbone.bottom_up.res4.4.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.5.conv1.*    | backbone.bottom_up.res4.5.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.5.conv2.*    | backbone.bottom_up.res4.5.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.5.conv3.*    | backbone.bottom_up.res4.5.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.6.conv1.*    | backbone.bottom_up.res4.6.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.6.conv2.*    | backbone.bottom_up.res4.6.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.6.conv3.*    | backbone.bottom_up.res4.6.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.7.conv1.*    | backbone.bottom_up.res4.7.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.7.conv2.*    | backbone.bottom_up.res4.7.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.7.conv3.*    | backbone.bottom_up.res4.7.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.8.conv1.*    | backbone.bottom_up.res4.8.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.8.conv2.*    | backbone.bottom_up.res4.8.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.8.conv3.*    | backbone.bottom_up.res4.8.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res4.9.conv1.*    | backbone.bottom_up.res4.9.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,1024,1,1)      |
| backbone.bottom_up.res4.9.conv2.*    | backbone.bottom_up.res4.9.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (256,) (256,) (256,) (256,) (256,256,3,3)       |
| backbone.bottom_up.res4.9.conv3.*    | backbone.bottom_up.res4.9.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (1024,) (1024,) (1024,) (1024,) (1024,256,1,1)  |
| backbone.bottom_up.res5.0.conv1.*    | backbone.bottom_up.res5.0.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,1024,1,1)      |
| backbone.bottom_up.res5.0.conv2.*    | backbone.bottom_up.res5.0.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.0.conv3.*    | backbone.bottom_up.res5.0.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.0.shortcut.* | backbone.bottom_up.res5.0.shortcut.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight} | (2048,) (2048,) (2048,) (2048,) (2048,1024,1,1) |
| backbone.bottom_up.res5.1.conv1.*    | backbone.bottom_up.res5.1.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.1.conv2.*    | backbone.bottom_up.res5.1.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.1.conv3.*    | backbone.bottom_up.res5.1.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.res5.2.conv1.*    | backbone.bottom_up.res5.2.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,2048,1,1)      |
| backbone.bottom_up.res5.2.conv2.*    | backbone.bottom_up.res5.2.conv2.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (512,) (512,) (512,) (512,) (512,512,3,3)       |
| backbone.bottom_up.res5.2.conv3.*    | backbone.bottom_up.res5.2.conv3.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}    | (2048,) (2048,) (2048,) (2048,) (2048,512,1,1)  |
| backbone.bottom_up.stem.conv1.*      | backbone.bottom_up.stem.conv1.{norm.bias,norm.running_mean,norm.running_var,norm.weight,weight}      | (64,) (64,) (64,) (64,) (64,3,7,7)              |
| backbone.fpn_lateral3.*              | backbone.fpn_lateral3.{bias,weight}                                                                  | (256,) (256,512,1,1)                            |
| backbone.fpn_lateral4.*              | backbone.fpn_lateral4.{bias,weight}                                                                  | (256,) (256,1024,1,1)                           |
| backbone.fpn_lateral5.*              | backbone.fpn_lateral5.{bias,weight}                                                                  | (256,) (256,2048,1,1)                           |
| backbone.fpn_output3.*               | backbone.fpn_output3.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output4.*               | backbone.fpn_output4.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.fpn_output5.*               | backbone.fpn_output5.{bias,weight}                                                                   | (256,) (256,256,3,3)                            |
| backbone.top_block.p6.*              | backbone.top_block.p6.{bias,weight}                                                                  | (256,) (256,2048,3,3)                           |
| backbone.top_block.p7.*              | backbone.top_block.p7.{bias,weight}                                                                  | (256,) (256,256,3,3)                            |
| head.bbox_pred.*                     | head.bbox_pred.{bias,weight}                                                                         | (36,) (36,256,3,3)                              |
| head.bbox_subnet.0.*                 | head.bbox_subnet.0.{bias,weight}                                                                     | (256,) (256,256,3,3)                            |
| head.bbox_subnet.2.*                 | head.bbox_subnet.2.{bias,weight}                                                                     | (256,) (256,256,3,3)                            |
| head.bbox_subnet.4.*                 | head.bbox_subnet.4.{bias,weight}                                                                     | (256,) (256,256,3,3)                            |
| head.bbox_subnet.6.*                 | head.bbox_subnet.6.{bias,weight}                                                                     | (256,) (256,256,3,3)                            |
| head.cls_score.*                     | head.cls_score.{bias,weight}                                                                         | (63,) (63,256,3,3)                              |
| head.cls_subnet.0.*                  | head.cls_subnet.0.{bias,weight}                                                                      | (256,) (256,256,3,3)                            |
| head.cls_subnet.2.*                  | head.cls_subnet.2.{bias,weight}                                                                      | (256,) (256,256,3,3)                            |
| head.cls_subnet.4.*                  | head.cls_subnet.4.{bias,weight}                                                                      | (256,) (256,256,3,3)                            |
| head.cls_subnet.6.*                  | head.cls_subnet.6.{bias,weight}                                                                      | (256,) (256,256,3,3)                            |

from detectron2.evaluation import COCOEvaluator, inference_on_dataset
from detectron2.data import build_detection_test_loader
from detectron2.engine import DefaultTrainer

evaluator = COCOEvaluator("rel_actions_val3", cfg, False, output_dir="./output_evals/")
val_loader = build_detection_test_loader(cfg, "rel_actions_val3")
inference_on_dataset(trainer.model, val_loader, evaluator)

WARNING [05/09 01:10:24 d2.evaluation.coco_evaluation]: COCO Evaluator instantiated using config, this is deprecated behavior. Please pass in explicit arguments instead.
[05/09 01:10:24 d2.data.datasets.coco]: Loaded 2215 images in COCO format from ./data/instances_vcoco_actions_val_2014.json
[05/09 01:10:24 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in inference: [ResizeShortestEdge(short_edge_length=(800, 800), max_size=1333, sample_style='choice')]
[05/09 01:10:24 d2.data.common]: Serializing 2215 elements to byte tensors and concatenating them all ...
[05/09 01:10:25 d2.data.common]: Serialized dataset takes 5.66 MiB
[05/09 01:10:25 d2.evaluation.evaluator]: Start inference on 2215 batches
[05/09 01:10:26 d2.evaluation.evaluator]: Inference done 11/2215. Dataloading: 0.0016 s/iter. Inference: 0.0780 s/iter. Eval: 0.0004 s/iter. Total: 0.0800 s/iter. ETA=0:02:56
[05/09 01:10:31 d2.evaluation.evaluator]: Inference done 69/2215. Dataloading: 0.0025 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0872 s/iter. ETA=0:03:07
[05/09 01:10:36 d2.evaluation.evaluator]: Inference done 127/2215. Dataloading: 0.0024 s/iter. Inference: 0.0839 s/iter. Eval: 0.0004 s/iter. Total: 0.0868 s/iter. ETA=0:03:01
[05/09 01:10:41 d2.evaluation.evaluator]: Inference done 186/2215. Dataloading: 0.0023 s/iter. Inference: 0.0837 s/iter. Eval: 0.0004 s/iter. Total: 0.0866 s/iter. ETA=0:02:55
[05/09 01:10:46 d2.evaluation.evaluator]: Inference done 244/2215. Dataloading: 0.0024 s/iter. Inference: 0.0838 s/iter. Eval: 0.0004 s/iter. Total: 0.0867 s/iter. ETA=0:02:50
[05/09 01:10:51 d2.evaluation.evaluator]: Inference done 301/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:02:46
[05/09 01:10:56 d2.evaluation.evaluator]: Inference done 359/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:02:41
[05/09 01:11:01 d2.evaluation.evaluator]: Inference done 417/2215. Dataloading: 0.0024 s/iter. Inference: 0.0840 s/iter. Eval: 0.0004 s/iter. Total: 0.0869 s/iter. ETA=0:02:36
[05/09 01:11:06 d2.evaluation.evaluator]: Inference done 474/2215. Dataloading: 0.0024 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0871 s/iter. ETA=0:02:31
[05/09 01:11:11 d2.evaluation.evaluator]: Inference done 532/2215. Dataloading: 0.0024 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0871 s/iter. ETA=0:02:26
[05/09 01:11:16 d2.evaluation.evaluator]: Inference done 591/2215. Dataloading: 0.0024 s/iter. Inference: 0.0840 s/iter. Eval: 0.0004 s/iter. Total: 0.0869 s/iter. ETA=0:02:21
[05/09 01:11:21 d2.evaluation.evaluator]: Inference done 650/2215. Dataloading: 0.0024 s/iter. Inference: 0.0839 s/iter. Eval: 0.0004 s/iter. Total: 0.0868 s/iter. ETA=0:02:15
[05/09 01:11:26 d2.evaluation.evaluator]: Inference done 707/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:02:11
[05/09 01:11:31 d2.evaluation.evaluator]: Inference done 765/2215. Dataloading: 0.0024 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:02:06
[05/09 01:11:36 d2.evaluation.evaluator]: Inference done 823/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:02:01
[05/09 01:11:41 d2.evaluation.evaluator]: Inference done 881/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:56
[05/09 01:11:47 d2.evaluation.evaluator]: Inference done 939/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:50
[05/09 01:11:52 d2.evaluation.evaluator]: Inference done 996/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:46
[05/09 01:11:57 d2.evaluation.evaluator]: Inference done 1055/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:40
[05/09 01:12:02 d2.evaluation.evaluator]: Inference done 1112/2215. Dataloading: 0.0024 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0871 s/iter. ETA=0:01:36
[05/09 01:12:07 d2.evaluation.evaluator]: Inference done 1170/2215. Dataloading: 0.0024 s/iter. Inference: 0.0841 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:30
[05/09 01:12:12 d2.evaluation.evaluator]: Inference done 1228/2215. Dataloading: 0.0024 s/iter. Inference: 0.0842 s/iter. Eval: 0.0004 s/iter. Total: 0.0870 s/iter. ETA=0:01:25
[05/09 01:12:17 d2.evaluation.evaluator]: Inference done 1281/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0876 s/iter. ETA=0:01:21
[05/09 01:12:22 d2.evaluation.evaluator]: Inference done 1339/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0876 s/iter. ETA=0:01:16
[05/09 01:12:27 d2.evaluation.evaluator]: Inference done 1397/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:01:11
[05/09 01:12:32 d2.evaluation.evaluator]: Inference done 1455/2215. Dataloading: 0.0024 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:01:06
[05/09 01:12:37 d2.evaluation.evaluator]: Inference done 1512/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:01:01
[05/09 01:12:42 d2.evaluation.evaluator]: Inference done 1569/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:00:56
[05/09 01:12:47 d2.evaluation.evaluator]: Inference done 1626/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0876 s/iter. ETA=0:00:51
[05/09 01:12:52 d2.evaluation.evaluator]: Inference done 1684/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0876 s/iter. ETA=0:00:46
[05/09 01:12:57 d2.evaluation.evaluator]: Inference done 1741/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0876 s/iter. ETA=0:00:41
[05/09 01:13:02 d2.evaluation.evaluator]: Inference done 1800/2215. Dataloading: 0.0024 s/iter. Inference: 0.0847 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:00:36
[05/09 01:13:07 d2.evaluation.evaluator]: Inference done 1859/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0875 s/iter. ETA=0:00:31
[05/09 01:13:13 d2.evaluation.evaluator]: Inference done 1918/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:25
[05/09 01:13:18 d2.evaluation.evaluator]: Inference done 1975/2215. Dataloading: 0.0024 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:20
[05/09 01:13:23 d2.evaluation.evaluator]: Inference done 2033/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:15
[05/09 01:13:28 d2.evaluation.evaluator]: Inference done 2090/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:10
[05/09 01:13:33 d2.evaluation.evaluator]: Inference done 2148/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:05
[05/09 01:13:38 d2.evaluation.evaluator]: Inference done 2206/2215. Dataloading: 0.0023 s/iter. Inference: 0.0846 s/iter. Eval: 0.0004 s/iter. Total: 0.0874 s/iter. ETA=0:00:00
[05/09 01:13:39 d2.evaluation.evaluator]: Total inference time: 0:03:13.212745 (0.087427 s / iter per device, on 1 devices)
[05/09 01:13:39 d2.evaluation.evaluator]: Total inference pure compute time: 0:03:06 (0.084569 s / iter per device, on 1 devices)
[05/09 01:13:40 d2.evaluation.coco_evaluation]: Preparing results for COCO format ...
[05/09 01:13:40 d2.evaluation.coco_evaluation]: Saving results to ./output_evals/coco_instances_results.json
[05/09 01:13:40 d2.evaluation.coco_evaluation]: Evaluating predictions with unofficial COCO API...
Loading and preparing results...
DONE (t=0.80s)
creating index...
index created!
[05/09 01:13:41 d2.evaluation.fast_eval_api]: Evaluate annotation type *bbox*
[05/09 01:13:43 d2.evaluation.fast_eval_api]: COCOeval_opt.evaluate() finished in 1.60 seconds.
[05/09 01:13:43 d2.evaluation.fast_eval_api]: Accumulating evaluation results...
[05/09 01:13:43 d2.evaluation.fast_eval_api]: COCOeval_opt.accumulate() finished in 0.45 seconds.
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.150
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.230
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.166
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.700
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.183
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.148
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.241
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.371
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.375
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.700
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.338
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.395
[05/09 01:13:43 d2.evaluation.coco_evaluation]: Evaluation results for bbox: 
|   AP   |  AP50  |  AP75  |  APs   |  APm   |  APl   |
|:------:|:------:|:------:|:------:|:------:|:------:|
| 14.973 | 22.975 | 16.646 | 70.000 | 18.309 | 14.828 |
[05/09 01:13:43 d2.evaluation.coco_evaluation]: Per-category bbox AP: 
| category      | AP     | category   | AP    | category   | AP     |
|:--------------|:-------|:-----------|:------|:-----------|:-------|
| hold          | 37.947 | carry      | 5.420 | point      | 0.000  |
| eat           | 12.513 | drink      | 0.000 | stand      | 48.934 |
| talk_on_phone | 0.000  |            |       |            |        |

OrderedDict([('bbox',
              {'AP': 14.973408472126575,
               'AP-carry': 5.420342633782121,
               'AP-drink': 0.0,
               'AP-eat': 12.512506011482463,
               'AP-hold': 37.9474266183727,
               'AP-point': 0.0,
               'AP-stand': 48.933584041248736,
               'AP-talk_on_phone': 0.0,
               'AP50': 22.974768706425476,
               'AP75': 16.64620894499706,
               'APl': 14.828050762899277,
               'APm': 18.308672603489068,
               'APs': 70.0})])

Run Action on Images

%ls ../GroceryImages

image01.jpeg  image07.jpeg  image13.jpeg  image40.jpeg  image46.jpeg
image02.jpeg  image08.jpeg  image14.jpeg  image41.jpeg  image47.jpeg
image03.jpeg  image09.png   image36.jpeg  image42.jpeg  image48.jpeg
image04.jpeg  image10.jpeg  image37.jpeg  image43.jpeg  image49.jpeg
image05.jpeg  image11.png   image38.jpeg  image44.jpeg  image50.jpeg
image06.png   image12.jpeg  image39.jpeg  image45.jpeg

image_list = ['image01.jpeg', 'image02.jpeg', 'image03.jpeg', 'image04.jpeg', 'image05.jpeg', 'image06.png', 'image07.jpeg', 'image08.jpeg', 'image09.png', 'image10.jpeg', 'image11.png', 'image12.jpeg', 'image13.jpeg', 'image14.jpeg', 'image36.jpeg', 'image37.jpeg', 'image38.jpeg', 'image39.jpeg', 'image40.jpeg', 'image41.jpeg', 'image42.jpeg', 'image43.jpeg', 'image44.jpeg', 'image45.jpeg', 'image46.jpeg', 'image47.jpeg', 'image48.jpeg', 'image49.jpeg', 'image50.jpeg']

for i in image_list:
  im = cv2.imread(f"../GroceryImages/{i}")
  print(i)
  outputs = predictor(im)
  v = Visualizer(im[:, :, ::-1], MetadataCatalog.get(cfg.DATASETS.TRAIN[0]), scale=1.2)
  out = v.draw_instance_predictions(outputs["instances"].to("cpu"))
  cv2_imshow(out.get_image()[:, :, ::-1])
  out.save(f"../sandbox/annotated_shopping/action_only_model/{i}")

image01.jpeg

image02.jpeg

image03.jpeg

image04.jpeg

image05.jpeg

image06.png